Search is not available for this dataset
text
stringlengths 1
129k
| id
stringlengths 9
16
| article_id
stringlengths 6
10
| section_title
stringlengths 1
1.26k
| educational_score
float64 0.46
5.16
| domain
stringclasses 3
values | document_type
stringclasses 4
values | domain_scores
sequencelengths 3
3
| document_type_scores
sequencelengths 4
4
| language
stringclasses 42
values | language_score
float64 0
1
|
|---|---|---|---|---|---|---|---|---|---|---|
About 42 (14.38 %) had 15 mg/kg to 80 mg/kg, 9 (3.08 %) had >80 mg/kg, 188 (64.4 %) had 1.1 mg/kg to 14.9 mg/kg, and 53 (18.2 %) had no iodine in the salt (0 mg/kg). Only 26 (8.9 %) of the households had used iodized salt properly. | PMC467049_p22 | PMC467049 | Results and discussion | 2.332361 | biomedical | Study | [
0.995439887046814,
0.0018366670701652765,
0.002723432146012783
] | [
0.9832947254180908,
0.01593725197017193,
0.00026275732670910656,
0.0005052314954809844
] | en | 0.857142 |
Generally, commercial salts have higher iodine levels compared to local salts, which agrees with the present study. The amount of iodine in salt from the studied villages was influenced by regulatory measures, processing methods , and the natural variability of source mineral deposits present, clearly observed in commercial-branded and local salt . Variability in iodine content of commercial-branded, SA (Kaysalt), and SB (Néel Premium) depends on the method used to fortify salt with iodine , transportation and distribution , and quality control measures during salt production , affecting the final iodine content. Proper handling and storage maintain the intended iodine levels in the salt . Exposure to moisture and sunlight affects iodine stability in salt degradation over time , as observed in local salts that were in an open state and not packed. The natural iodine content of the raw salt source may result in some salt deposits naturally containing more iodine than others experienced in the studied villages. | PMC467049_p23 | PMC467049 | Results and discussion | 4.070018 | biomedical | Study | [
0.997081458568573,
0.00022196849749889225,
0.0026966154109686613
] | [
0.9991106390953064,
0.00046340422704815865,
0.0003878129064105451,
0.000038125825085444376
] | en | 0.999997 |
Results in Table 3 indicate that most samples meet the regulatory standards set by TBS and WHO for nitrate and sulphate content, with the exception of phosphate. The concentration range was nitrate (3.30–4.40 mg/kg, 5.45–7.40 mg/kg), sulphate (0.31–0.42 mg/kg, 0.03–0.07 mg/kg) for local and commercially branded salt, respectively. Phosphate was not detected in commercially branded salt; for local salt, the amount ranged from 0.02 mg/kg to 0.17 mg/kg. There were significant differences observed between the salt nitrate, phosphate, and sulphate contents of the five villages (p < 0.001). Results in Table 3 and 4 suggest variations in nitrate, phosphate, sulphate, ammonia, copper, iron, and manganese concentrations in local and commercially branded salt samples, suggesting potential differences in production methods , sourcing , and quality control measures . Disparities in mineral content between commercial and local salt have implications for observed nutritional adequacy in Mangwanjuki (sample SB), Chibumagwa (sample SD), and Chali Igongo (sample SE) villages, affecting areas where salt is a primary source of these minerals. The consistency of some cations and anions concentrations across different sources of salt observed indicates uniformity in mineral composition and quality assurance practices in commercially branded salts. Table 3 Amount of nitrate, phosphate, and sulphate available in commercial-branded and local salt. Table 3 Sample Nitrate (mg/Kg) ± SD, n = 3 Phosphate (mg/Kg) ± SD, n = 3 Sulphate (mg/Kg) ± SD, n = 3 p - value SA 4.40 ± 0.20 0.11 ± 0.01 0.42 ± 0.02 <0.001 SB 3.30 ± 0.10 0.13 ± 0.02 0.37 ± 0.02 <0.001 SC 3.50 ± 0.10 0.17 ± 0.01 0.31 ± 0.01 <0.001 SD 4.10 ± 0.10 0.02 ± 0.01 0.39 ± 0.01 <0.001 SE 4.40 ± 0.20 0.48 ± 0.02 0.34 ± 0.01 <0.001 SF 5.45 ± 0.10 ND 0.03 ± 0.01 <0.001 SG 7.40 ± 0.10 ND 0.07 ± 0.01 <0.001 TBS, 2014 5.0 00 0.5 WHO, 2018 45.0 0.1 1.0 | PMC467049_p24 | PMC467049 | Results and discussion | 4.166809 | biomedical | Study | [
0.9980968832969666,
0.0003475828270893544,
0.001555604045279324
] | [
0.9995304346084595,
0.00017317140009254217,
0.00025788048515096307,
0.000038595200749114156
] | en | 0.999997 |
According to Table 4 , the ammonia content was consistent across all samples, with values ranging from 0.5 mg/kg to 0.6 mg/kg falling within the range recommended (1.0 mg/kg) and found to be not significant statistically (p > 0.001). Copper concentrations ranged from 0.9 mg/kg to 2.0 mg/kg across the samples, with samples SA and SB displaying the highest values. Iron levels were between 0.5 mg/kg and 1.8 mg/kg, within the TBS (5.0 mg/kg) and WHO (0.1 mg/kg) limits. Manganese concentrations ranged from 0.5 mg/kg to 1.8 mg/kg within regulatory standards (1.0 mg/kg). The concentrations were statistically significant (p < 0.001) for copper, iron, and manganese. Table 4 Concentration of ammonia, copper, iron, and manganese in salt. Table 4 Sample Ammonia (mg/Kg) ± SD, (n = 3) Copper (mg/Kg) ± SD, (n = 3) Iron (mg/Kg) ± SD, (n = 3) Manganese (mg/Kg) ± SD, (n = 3) SA 0.5 ± 0.1 2.0 ± 0.2 1.8 ± 0.1 1.8 ± 0.1 SB 0.5 ± 0.1 2.0 ± 0.2 1.0 ± 0.1 1.0 ± 0.1 SC 0.5 ± 0.1 1.0 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 SD 0.5 ± 0.1 1.0 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 SE 0.5 ± 0.1 1.0 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 SF 0.6 ± 0.2 0.9 ± 0.1 0.9 ± 0.1 0.9 ± 0.1 SG 0.5 ± 0.1 1.0 ± 0.1 0.9 ± 0.1 0.9 ± 0.1 p value p > 0.001 p < 0.001 p < 0.001 p < 0.001 TBS, 2014 1.0 2.0 5.0 1.0 WHO, 2018 1.0 0.1 1.0 1.0 | PMC467049_p25 | PMC467049 | Results and discussion | 4.159823 | biomedical | Study | [
0.9985900521278381,
0.0003186814137734473,
0.00109126849565655
] | [
0.9995594620704651,
0.00015548816008958966,
0.00024770444724708796,
0.0000373163384210784
] | en | 0.999998 |
In a study by Ref. on metal contaminants Tehran in edible salts, Pb (1.59 ± 0.90), Cd (0.91 ± 0.32), Zn (6.02 ± 2.54), Fe (17.8 ± 6.11), Cu (1.24 ± 0.90), and Al (5.82 ± 0.61) mg/Kg in salts were Pb (0.86 ± 0.52), Cd (0.65 ± 0.34), Zn (6.5 ± 4.86), Fe (15.3 ± 5.95), Cu (1.21 ± 0.79), and Al (5.60 ± 0.75) mg/Kg in table salts . study on the mineral composition showed that the composition of copper (Cu) in Mozia and Persian blue salt was 41.63 ± 4.18 mg/kg and 50.61 ± 6.19 mg/kg, respectively. Iron (Fe) 1.44 ± 0.26 mg/kg had the highest manganese (Mn) content with 5.15 ± 0.62 mg/kg while Maldon salt had the lowest with 0.82 ± 0.04 mg/kg. Comparing these results, there are consistent mineral content differences between commercial and local salts. Non-compliance with mineral content requirements in salt has significant health implications, including nutritional deficiencies , increased health risks , population health burdens , and vulnerability to certain demographic groups . Inadequate mineral intake leads to osteoporosis , cardiovascular diseases , and muscle cramps , and legal and regulatory consequences for manufacturers and producers, such as recalls and fines , undermine consumer confidence in food products' safety and quality , leading to skepticism about fortification programs . According to Ref. , addressing non-compliance requires targeted interventions, improved production practices, enhanced monitoring, and public education campaigns; therefore, collaboration between government agencies, industry stakeholders, and public health organizations is essential to ensuring compliance. | PMC467049_p26 | PMC467049 | Results and discussion | 4.177166 | biomedical | Study | [
0.9989134073257446,
0.00023131218040362,
0.0008554019150324166
] | [
0.9823974370956421,
0.0005199023289605975,
0.016989009454846382,
0.00009367585153086111
] | en | 0.999996 |
The socioeconomic characteristics of households in Nkonkilangi, Mangwanjuki, Unyanga, Chibumagwa, and Kinyambwa villages are presented in Table 5 , categorized as education level, income, and occupation. Table 5 Household socioeconomic characteristics. Table 5 Socioeconomic status Category Villages Nkonkilangi Mangwanjuki Unyanga Chibumagwa Kinyambwa Education Informal 13 (08.0 %) 09 (9.4 %) 11 (20.4 %) 24 (22.6 %) 23 (45.1 %) Primary 74 (45.4 %) 27 (28.1 %) 08 (14.8 %) 41 (38.7 %) 17 (33.3 %) Secondary 51 (31.3 %) 31 (32.3 %) 22 (40.7 %) 33 (31.1 %) 07 (13.7 %) Tertiary 25 (15.3 %) 29 (30.2 %) 13 (24.1 %) 08 (07.6 %) 04 (7.8 %) 163 (100 %) 96 (100 %) 54 (100 %) 106 (100 %) 51(100 %) Income Low 88 (54.0 %) 27 (28.1 %) 29 (53.7 %) 33 (31.1 %) 15 (29.4 %) Middle 63 (38.6 %) 52 (54.2 %) 21 (38.9 %) 46 (43.4 %) 27 (52.9 %) High 12 (07.4 %) 17 (17.7 %) 04 (7.4 %) 27 (25.5 %) 09 (17.7 %) 163(100 %) 96 (100 %) 54 (100 %) 106 (100 %) 51 (100 %) Occupation Peasant 103 (63.2 %) 57 (59.4 %) 34 (61.8 %) 45 (42.4 %) 27 (53.0 %) Business 23 (14.1 %) 18 (18.8 %) 11 (20 %) 39 (36.8 %) 12 (23.5 %) Employee 37 (22.7 %) 21 (21.8 %) 10 (18.2 %) 22 (20.8 %) 12 (23.5 %) 163(100 %) 96 (100 %) 54 (100 %) 106 (100 %) 51 (100 %) | PMC467049_p27 | PMC467049 | Results and discussion | 3.111598 | biomedical | Study | [
0.624915361404419,
0.0010617506923153996,
0.37402284145355225
] | [
0.9963648915290833,
0.003295073052868247,
0.0002702494093682617,
0.00006988727545831352
] | en | 0.999998 |
The highest percentage of households with informal education was found in Chibumagwa 24 (22.6 %), followed by Kinyambwa 23 (45.1 %). Nkonkilangi village has the highest percentage of households with primary and secondary education, at 74 (45.4 %) and 51 (31.3 %), respectively. | PMC467049_p28 | PMC467049 | Results and discussion | 1.421987 | other | Other | [
0.12693656980991364,
0.000982103869318962,
0.8720813393592834
] | [
0.28051555156707764,
0.7175818681716919,
0.0009718510555103421,
0.0009307694854214787
] | en | 0.999997 |
Tertiary education is least prevalent across all villages, with the highest percentage of 29 (30.2 %) in Mangwanjuki village. As for income levels, Nkonkilangi has the highest percentage of households with low income at 88 (54.0 %), followed by Chibumagwa at 33 (31.1 %). This suggests a significant proportion of households in these villages face financial challenges. Mangwanjuki and Kinyambwa have the highest percentages of households with middle income: 52 (54.2 %) and 27 (52.9 %), respectively. Chibumagwa and Mangwanjuki have the highest percentages of households with high income: 27 (25.5 %) and 17 (17.7 %), respectively. The majority of households across all villages were peasants, with percentages ranging from 45 (42.4 %) to 103 (63.2 %). Chibumagwa has the highest number of households doing business, at 39 (36.8 %), followed by Kinyambwa at 12 (23.5 %). Households engaged in employment ranged from 10 (18.2 %) to 12 (23.5 %) across villages. Variation in education levels across villages, with some showing higher proportions of informal and primary education shown in Table 5 , is likely to affect choice of salt due to a lack of knowledge on salt quality . The majority of households across all villages fall into the low- and middle-income categories. Low- and middle-income households are more sensitive to prices, opting for cheaper options, which could include locally produced salt, which is more affordable compared to branded salt . In many rural areas, locally produced salt might be more readily available than branded salt in purchase decisions, especially when transportation costs are factored in Refs. . According to the interview, locally produced salt was perceived as more natural and authentic by some consumers. This perception could influence their choice, as they value traditional methods and are more concerned about additives in commercial products. On the other hand, middle-income households, in particular, were more aware of the nutritional content of salt and its health implications. This awareness led them to choose branded salt that is iodized for their family health. Low- and middle-income households considered value for money as branded salt was more expensive and opted to use local salt. | PMC467049_p29 | PMC467049 | Results and discussion | 2.64523 | other | Study | [
0.4027791917324066,
0.0009917316492646933,
0.596229076385498
] | [
0.9866993427276611,
0.01280552614480257,
0.00034688020241446793,
0.00014820574142504483
] | en | 0.999997 |
It is indicated in Table 5 that peasants dominated in all villages, forming the majority of households with the highest percentages. While price, quality, accessibility, and convenience are important factors for all groups, the specific priorities and considerations of peasants, businessmen, and employees led them to make different choices when it comes to local salt versus commercial salt. Businessmen prioritized convenience and time-saving options, opting for branded salt available in shops due to its accessibility and the convenience of purchasing multiple items in one location. Businessmen trusted branded salt more due to standardized production processes and quality control measures, leading them to choose it over local salt, which also associated branded products with their lifestyle and status. Employees with stable incomes prioritized health and nutrition by choosing branded salt, which is often seen as a healthier option for themselves and their families. In addition, employees have brand loyalty, preferring to stick to familiar brands that they trust and choosing branded salt because they have positive associations with the brand from previous experiences. These variations in salt consumption patterns were influenced by availability, affordability, cultural practices, and perceptions of salt quality. These consumption patterns can be used to design targeted interventions aimed at promoting the consumption of adequately iodized salt within these communities. | PMC467049_p30 | PMC467049 | Results and discussion | 2.273582 | other | Study | [
0.33379003405570984,
0.0007624050485901535,
0.6654475331306458
] | [
0.9640594124794006,
0.03497982397675514,
0.0007466640672646463,
0.0002140205615432933
] | en | 0.999998 |
Table 6 shows household consumption of commercial-branded and local salt in different sources and villages, in which out of 470 households surveyed, 234 (49.88 %) consumed local and 236 (50.12 %) consumed commercial-branded salt, with varying prevalence of the types of salts across villages that were significantly different (p < 0.001) across different villages. Mangwanjuki village recorded the highest preference for commercial-branded salt (80 (82.8 %), whereas Chali Igongo village predominantly consumes local salt (38 (74.6 %). Other villages exhibited a mix of varying preferences for both commercial-branded and local salt. Results suggest varying preferences for salt consumption across different villages, with some showing a clear preference for local salt while others overwhelmingly prefer commercial salt. The observed variations likely result from a combination of these factors, reflecting the complex interplay of cultural, economic, geographic, and social influences on dietary choices within each village. Different villages had distinct cultural preferences regarding food, including the type of salt used in cooking and seasoning, stemming from historical practices and beliefs about the health benefits of certain types of salt. Economic considerations, such as the affordability and availability of commercial-branded versus local salt, influenced consumption patterns. Commercial-branded salt was more easily accessible in urban areas such as Mangwanjuki and Unyanga villages, leading to higher consumption rates. Access to markets and salt distribution varied between villages, impacting the availability of different types of salt. Villages with better access to markets had a wider variety of salt options, including commercial-branded and local varieties, leading to more balanced consumption patterns. Risks associated with different types of salt raised concerns about additives in commercial-branded salt, while others opted for commercial-branded salt based on perceived quality and safety standards. Villages located near salt production sites had easier access to local salt, while those farther away relied more heavily on commercially-produced varieties. Table 6 Household consumption of commercially-branded and local salt. Table 6 Village Local salt % Commercial salt % Total households p - value Nkonkilangi 110 (67.9 %) 53 (32.1 %) 163 <0.001 Mangwanjuki 16 (17.2 %) 80 (82.8 %) 96 <0.001 Unyanga 15 (26.7 %) 39 (73.3 %) 54 <0.001 Chibumagwa 67 (63.0 %) 39 (37.0 %) 106 <0.001 Chali Igongo 38 (74.6 %) 12 (25.4 %) 51 <0.001 234 (49.88 %) 236 (50.12 %) 470 | PMC467049_p31 | PMC467049 | Results and discussion | 3.968595 | biomedical | Study | [
0.8760861158370972,
0.0008269117097370327,
0.12308693677186966
] | [
0.9989891648292542,
0.0006026591290719807,
0.00036663480568677187,
0.00004149301821598783
] | en | 0.999998 |
The study revealed a significant difference in the composition of minerals in commercially branded versus locally sourced salts from Bahi, Iramba, Manyoni, and Singida urban districts in Tanzania. Local salts often had higher concentrations of essential minerals compared to branded salts, yet iodine content was inconsistent and sometimes below recommended levels. These findings call the need for improved regulation and standardization of salt production to ensure consistent mineral intake and prevent deficiencies, particularly of iodine, which is important for thyroid health. The limitations of the study include a relatively small sample size and a limited geographic focus, which may not fully capture the variability in mineral content across different regions. In addition, did not account for potential seasonal variations in mineral content. Future research should broaden the sampling area, and consider seasonal effects. This study calls the need for standardized regulations to ensure consistent edible salt quality and consumer awareness for health benefits. | PMC467049_p32 | PMC467049 | Conclusions | 4.115186 | biomedical | Study | [
0.9987251162528992,
0.00026701611932367086,
0.0010078640189021826
] | [
0.9993522763252258,
0.0001861991040641442,
0.0004172396438661963,
0.00004430569970281795
] | en | 0.999998 |
The data will be made available upon request. | PMC467049_p33 | PMC467049 | Data availability statement | 0.769031 | other | Other | [
0.22128818929195404,
0.006519731599837542,
0.772192120552063
] | [
0.01288018748164177,
0.9837489128112793,
0.001982127083465457,
0.001388727338053286
] | fr | 0.714281 |
Questionnaire survey for households in the studied villages. | PMC467049_p34 | PMC467049 | Additional information | 1.525435 | biomedical | Other | [
0.7665025591850281,
0.0030863862484693527,
0.230411097407341
] | [
0.21204185485839844,
0.7843993902206421,
0.0022994973696768284,
0.0012592645362019539
] | en | 0.999996 |
Jackson Henry Katonge: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. | PMC467049_p35 | PMC467049 | CRediT authorship contribution statement | 1.008929 | other | Other | [
0.31934747099876404,
0.005321097560226917,
0.6753314733505249
] | [
0.0068001458421349525,
0.9921900629997253,
0.00046145013766363263,
0.0005483854911290109
] | en | 0.999995 |
The author declares that have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. | PMC467049_p36 | PMC467049 | Declaration of competing interest | 0.982726 | other | Other | [
0.00418325699865818,
0.0007736874977126718,
0.9950430393218994
] | [
0.0014122514985501766,
0.9971388578414917,
0.0007751026423648,
0.0006737741641700268
] | en | 0.999997 |
With the acceleration of people's lifestyles and the rise of emerging industries, such as logistics and food delivery, electric bicycles and their convenience, environmental friendliness, and energy efficiency, have become the preferred mode of urban transportation. However, in complex traffic environments, the failure of e-bike riders to wear safety helmets poses a constant threat to personal safety. Research indicates that approximately 80 % of annual accidents, including motorcycle and e-bike riders, are caused by the absence of safety helmets. Despite the nationwide 'One Helmet, One Belt' safety campaign, the limited awareness of safety among riders, low penalties for violations coupled with the constraints of limited law enforcement resources, inefficient on-site helmet compliance checks, and high costs, hinder a substantial increase in safety helmet usage rates among e-bike riders. Therefore, using artificial intelligence technology for detecting whether e-bike riders wear helmets and assisting traffic enforcement personnel in tracking and identifying violations is of significant importance for road traffic safety. | PMC467071_p0 | PMC467071 | Introduction | 1.897356 | other | Other | [
0.023298390209674835,
0.00036966503830626607,
0.9763319492340088
] | [
0.14988476037979126,
0.8427413702011108,
0.006606808863580227,
0.0007670148625038564
] | en | 0.999998 |
In recent years, object tracking algorithms in computer vision technology have made significant advancements and yielded fruitful results. In 2016, Bertinetto et al. introduced SiamFC, which is a single-object tracking algorithm based on Siamese networks. It employs an end-to-end fully convolutional network for estimating object positions through feature extraction and cross-correlation computations. Building upon this foundation, additional single-object tracking algorithms, such as CFNet that combines correlation filtering and SiamMa that integrates semantic segmentation, have been proposed. These advancements have further enhanced tracking performance. Similarly, researchers have made significant contributions for object tracking algorithms. From the early DSST algorithm to subsequent innovations like CCOT , ECO , ATOM , and Dimp50 algorithms, the introduction of these algorithms has further propelled the development of the object tracking field. With the rapid evolution of object detectors, an increasing number of object tracking algorithms are increasing powerful object detectors to enhance tracking performance , such as the YOLO series of object detectors , YOLOR, YOLOX, Faster R–CNN , and others. | PMC467071_p1 | PMC467071 | Introduction | 2.644715 | other | Review | [
0.03285468369722366,
0.0007904766825959086,
0.9663547873497009
] | [
0.1335524618625641,
0.3236682415008545,
0.5406564474105835,
0.0021229020785540342
] | en | 0.999996 |
In this regard, Bewley et al. introduced an important multi-object tracking algorithm in 2016, namely the SORT algorithm. The algorithm utilizes Faster R–CNN as the object detector, employs Kalman filters for object trajectory prediction, and utilizes the Hungarian algorithm to find the specific matching object bounding box with the maximum IOU (Intersection Over Union) to achieve object matching between consecutive frames. Although the algorithm has the advantages of a simple structure and fast execution speed, it faces challenges in complex scenarios, such as unstable object ID associations and frequent ID switches. Thereafter, the DeepSORT algorithm was introduced based on SORT . This algorithm incorporates cascade matching and feature re-identification networks. Cascade matching enhances the tracking of occluded objects, while the feature re-identification network aids in distinguishing differences between different objects, thus improving the accuracy and robustness of multi-object tracking. | PMC467071_p2 | PMC467071 | Introduction | 3.25731 | other | Study | [
0.12203892320394516,
0.000560268119443208,
0.877400815486908
] | [
0.7509301900863647,
0.18810850381851196,
0.059833526611328125,
0.0011277742451056838
] | en | 0.999998 |
These improvements enable the DeepSORT algorithm to better handle object association challenges, especially in cases of prolonged occlusion, resulting in a significant enhancement in tracking performance. However, the SORT and DeepSORT algorithms often rely on bounding boxes with high confidence scores above specific thresholds for association while discarding them with low confidence scores due to issues, such as occlusion or motion blur. This has led to significant problems of object loss and track fragmentation that cannot be ignored. To address this issue, the ByteTrack algorithm employs a BYTE association method. It not only associates high-confidence bounding boxes but it also considers matching low-confidence bounding boxes to a certain extent, thereby enhancing accuracy of tracking. By utilizing YOLOX as the object detector and the BYTE association method, the ByteTrack algorithm can deliver higher performance in multi-object tracking tasks, thereby enhancing accuracy and robustness of tracking. | PMC467071_p3 | PMC467071 | Introduction | 2.24771 | other | Other | [
0.04991869628429413,
0.00045042377314530313,
0.949630856513977
] | [
0.32744094729423523,
0.6688194870948792,
0.002707431325688958,
0.0010321951704099774
] | en | 0.999996 |
Due to the exceptional performance of the Transformer architecture across various domains, researchers have been prompted to introduce it for multi-object tracking . Accordingly, in 2022, Zhou X et al. proposed an innovative global multi-object tracking algorithm based on the Transformer architecture. This approach took short video sequences as input, utilizing the Transformer to encode object features and generated specific trajectories, all without the need for intermediate pairwise grouping or combinatorial association. Furthermore, it could be jointly trained with an object detector, thereby enhancing the performance of global multi-object tracking. | PMC467071_p4 | PMC467071 | Introduction | 1.458394 | other | Other | [
0.023900968953967094,
0.0003520940663293004,
0.9757469892501831
] | [
0.36236920952796936,
0.6321194767951965,
0.0041100331582129,
0.001401263871230185
] | en | 0.999999 |
Subsequently, in 2023, Zhang Y et al. , building upon MOTR , introduced MOTRv2. They incorporated an additional object detector to provide prior detection information to MOTR, consequently enhancing end-to-end multi-object tracking performance. However, it should be noted that tracking algorithms based on the Transformer architecture faced challenges in terms of terminal deployment and runtime speed, necessitating further optimization. | PMC467071_p5 | PMC467071 | Introduction | 1.35796 | other | Other | [
0.032715193927288055,
0.0004527720157057047,
0.9668320417404175
] | [
0.1840580254793167,
0.8120114207267761,
0.0028229346498847008,
0.0011076867813244462
] | en | 0.999997 |
Currently, in research related to applications in traffic monitoring scenarios, Zihan P et al. conducted a study on the recognition of wrong-way riding behavior of electric bicycles. They employed a hybrid Gaussian model to extract background and used background subtraction to extract the foreground of electric bicycles. Furthermore, by combining Kalman filtering and vehicle centroid features, they predicted and tracked the characteristics of vehicles for the next moment. Ultimately, by analyzing changes in the centroid coordinates of electric bicycles, they successfully identified instances of wrong-way riding behavior. | PMC467071_p6 | PMC467071 | Introduction | 1.570218 | other | Study | [
0.03663725033402443,
0.0005315597518347204,
0.9628311991691589
] | [
0.9763830900192261,
0.022107068449258804,
0.0010594985214993358,
0.0004503388190641999
] | en | 0.999998 |
Xiaoping W et al. , in response to challenges related to occlusion, rotation, and scale transformation, have enhanced the MDnet algorithm to improve object tracking accuracy in complex traffic scenarios. They employed optical flow change information and small convolutional kernels to track and predict motor vehicles, non-motorized vehicles, and pedestrians. | PMC467071_p7 | PMC467071 | Introduction | 1.359434 | other | Study | [
0.02879306860268116,
0.00044711833470501006,
0.9707597494125366
] | [
0.5631974935531616,
0.4314024746417999,
0.003950160462409258,
0.0014498495729640126
] | en | 0.999996 |
Zhenxiao L et al. proposed a multi-vehicle tracking algorithm to address real-time performance and ID switching issues in multi-object tracking. They first employed YOLOv3 as the object detector and then, in combination with the DeepSORT tracking algorithm, introduced an LSTM motion model for tracking vehicle objects in traffic scenarios. | PMC467071_p8 | PMC467071 | Introduction | 1.405233 | other | Other | [
0.024295445531606674,
0.00038702672463841736,
0.9753174781799316
] | [
0.4333614408969879,
0.561896800994873,
0.003346176352351904,
0.0013955418253317475
] | en | 0.999998 |
Caihong L et al. have designed a cross-view multi-object tracking visualization algorithm based on field-of-view stitching. They utilized the geometric information of video scenes to achieve field-of-view stitching, presenting tracked objects from different perspectives in a unified field of view. This algorithm facilitated the presentation of surveillance scene information from multiple perspectives within the same field of view, providing a more convenient way to monitor traffic intersections. | PMC467071_p9 | PMC467071 | Introduction | 1.427235 | other | Other | [
0.026278361678123474,
0.00036938663106411695,
0.973352313041687
] | [
0.328293114900589,
0.6677243113517761,
0.002824708353728056,
0.0011578149860724807
] | en | 0.999997 |
The following table is a comprehensive overview of the literature review, outlining the strengths and weaknesses associated with each method. | PMC467071_p10 | PMC467071 | Introduction | 2.674774 | biomedical | Review | [
0.9505624175071716,
0.0059352898970246315,
0.04350230470299721
] | [
0.01703953556716442,
0.03400733694434166,
0.9478658437728882,
0.0010872252751141787
] | en | 0.999996 |
Precisely, the existing algorithms have not effectively addressed different issues, such as frequent ID switching and fragmented tracking trajectories in complex traffic scenarios. Electric bicycle riders may pass through the monitoring area from a distance or up close, leading to variable object scales. Furthermore, due to factors several like crowded areas and mutual occlusion, traditional object tracking algorithms often struggle to handle these complex traffic scenarios effectively, resulting in frequent ID switches, reduced tracking range and accuracy, and fragmented tracking trajectories. Therefore, achieving effective recognition of violations by electric bicycle riders necessitates the accurate differentiation of various objects and the continuous tracking of their behavioral trajectories. | PMC467071_p11 | PMC467071 | Introduction | 1.159822 | other | Other | [
0.006934342440217733,
0.0004448879917617887,
0.9926207065582275
] | [
0.04780033975839615,
0.9496470093727112,
0.0015925896586850286,
0.0009601233177818358
] | en | 0.999998 |
Despite the progress made in object tracking algorithms, there are still several gaps that require attention. Firstly, the current algorithms have not adequately addressed the challenge of frequent ID switching in complex traffic scenarios, resulting in fragmented tracking trajectories. This problem is especially prevalent in situations with high pedestrian density and frequent occlusions. Secondly, traditional object tracking algorithms often face difficulties in handling objects with varying scales, as they may appear in the area being monitoring from a distance or in close proximity. Lastly, there is a need to enhance the real-time performance and accuracy of object tracking algorithms to meet the practical demands of traffic monitoring scenarios. | PMC467071_p12 | PMC467071 | Introduction | 1.000967 | other | Other | [
0.0023589334450662136,
0.0004537659988272935,
0.9971873164176941
] | [
0.021497728303074837,
0.974371075630188,
0.0030304216779768467,
0.0011007458670064807
] | en | 0.999997 |
Therefore, in this paper, an electric bicycle tracking algorithm has been proposed, EBTrack, designed for traffic monitoring scenarios. | PMC467071_p13 | PMC467071 | Introduction | 0.996279 | other | Other | [
0.007405291311442852,
0.00045868507004342973,
0.9921361207962036
] | [
0.13112276792526245,
0.8658882975578308,
0.001368747791275382,
0.0016202219994738698
] | en | 0.999998 |
The EBTrack algorithm utilizes cutting-edge distance measurement technologies, as outlined by Liu and Bao , that make use of ultra-wideband sensors to enable accurate real-time monitoring of electric bicycles in city traffic. Additionally, our strategy builds upon the fundamental research on distance measurement technologies based on electromagnetic waves, as examined by Liu and Bao , which form the basis for the remote sensing capabilities crucial to the effectiveness of EBTrack. The lightweight YOLOv7 has been used as the object detector and ResNetEB has been introduced as the feature extraction network. An adaptive modulated noise scale Kalman filter has been also incorporated, and the association matching mechanism has been redesigned. | PMC467071_p14 | PMC467071 | Introduction | 1.866426 | other | Other | [
0.08057554066181183,
0.0004937835619784892,
0.9189306497573853
] | [
0.32323914766311646,
0.673678457736969,
0.0020361386705189943,
0.0010461615165695548
] | en | 0.999997 |
The EBTrack tracking algorithm offers several advantages. Firstly, by utilizing the lightweight YOLOv7 object detector, efficient and accurate object detection has been achieved, providing reliable object position information, thereby enhancing tracking algorithm accuracy and stability. Secondly, the ResNetEB feature extraction network structure has been introduced, specifically designed for feature extraction and re-identification of electric bicycles. This improves the performance of the tracking algorithm in complex scenarios with high pedestrian density and occlusion, providing a more reliable feature representation. Thirdly, the introduction of the adaptive modulated noise scale Kalman filter enhances the accuracy and stability of object trajectories. This allows the tracking algorithm to better adapt to changes in object motion and the environment. Finally, through the redesigned association matching mechanism, the issue of object ID switching has been successfully reduced, improving tracking stability and continuity while overcoming the problem of fragmented tracking trajectories. This mechanism brings the EBTrack algorithm more closely with practical requirements and establishes an accurate data foundation for effective violation recognition. | PMC467071_p15 | PMC467071 | Introduction | 2.751775 | other | Other | [
0.07264915853738785,
0.0005140670691616833,
0.9268367886543274
] | [
0.38986867666244507,
0.6045481562614441,
0.00477578304708004,
0.0008073496865108609
] | en | 0.999997 |
The proposed method presents a number of enhancements compared to the standard YOLOv7 model for electric bicycle tracking. The improvements include the introduction of the ResNetEB Feature Extraction Network, which is specifically designed for electric bicycle re-identification. This network enhances performance in scenarios with high pedestrian density and occlusion, providing more reliable feature representation. Additionally, an Adaptive Modulated Noise Scale Kalman Filter is incorporated to improve the accuracy and stability of object trajectories by adapting to changes in object motion and the environment. The Association Matching Mechanism is also redesigned to consider the special motion patterns of electric bicycles, reducing object ID switching and improving tracking stability. Furthermore, the algorithm is optimized for real-time performance, with feature extraction only occurring when generating new object IDs and the matching mechanism balanced for accuracy and speed. These enhancements enable the EBTrack algorithm to achieve superior performance in electric bicycle tracking tasks, enhancing data reliability for effective violation recognition. The primary contributions are outlined as follows. - Utilizing the lightweight YOLOv7 as the object detector has enabled the stakeholders to achieve efficient and accurate object detection, thereby providing dependable input data for subsequent tracking processes. - The introduction of the ResNetEB feature extraction network structure, specifically custom-made for feature extraction and re-identification of electric bicycles, has significantly enhanced the performance of the tracking algorithm in challenging scenarios characterized by high pedestrian density and occlusion. - The incorporation of an adaptive modulated noise scale Kalman filter has played a crucial role in improving the accuracy and stability of object trajectories. This enhancement allows the algorithm to effectively adapt to variations in object motion and environmental conditions. - The redesign of the association matching mechanism has successfully alleviated the issue of object ID switching, leading to improved tracking stability and continuity. Additionally, this modification has addressed the problem of fragmented tracking trajectories. | PMC467071_p16 | PMC467071 | Introduction | 2.612352 | other | Study | [
0.042733971029520035,
0.0004688746703322977,
0.9567971229553223
] | [
0.8290511965751648,
0.1666594296693802,
0.0034990946296602488,
0.0007902781362645328
] | en | 0.999996 |
The EBTrack tracking algorithm has established a more robust data foundation, thereby facilitating effective violation recognition in traffic monitoring scenarios. | PMC467071_p17 | PMC467071 | Introduction | 1.135092 | other | Other | [
0.014470815658569336,
0.0008158559212461114,
0.9847133755683899
] | [
0.0029062428511679173,
0.9961632490158081,
0.0005584899918176234,
0.00037209433503448963
] | en | 0.999997 |
YOLOv7 combines high accuracy with lightweight design, performing excellently across a range from 5FPS to 160FPS, making it a powerful solution for various application scenarios. Compared to previous detectors, such as YOLOX, SSD , and Faster R–CNN, YOLOv7 has achieved significant improvements in both accuracy and speed. These enhancements are attributed to the adoption of several key technologies, including the introduction of model reparameterization into the network architecture, the utilization of cross-grid label assignment technique of YOLOv5, and matching technique of YOLOX. | PMC467071_p18 | PMC467071 | The object detector YOLOv7 | 1.531159 | other | Other | [
0.0519154854118824,
0.0007100102957338095,
0.9473745226860046
] | [
0.020880499854683876,
0.9771003723144531,
0.0014590658247470856,
0.0005600018193945289
] | en | 0.999997 |
Additionally, YOLOv7 introduces a novel ELAN (Efficient Layer Aggregation Networks) structure, which maintains model performance while reducing computational complexity. As shown in Fig. 1 , where k represents convolution's kernel size, and s represents stride. YOLOv7 also introduces a training method with auxiliary heads, which enhances detection accuracy by increasing the training cost without affecting inference time. Fig. 1 Elan structure in YOLOv7. Fig. 1 | PMC467071_p19 | PMC467071 | The object detector YOLOv7 | 1.607971 | other | Other | [
0.13186083734035492,
0.0008660625899210572,
0.8672730922698975
] | [
0.07558800280094147,
0.9223964214324951,
0.001204240950755775,
0.0008112209034152329
] | en | 0.999997 |
The ByteTrack tracking algorithm is based on object detection. Similar to other non-Re-ID algorithms, it solely uses the bounding boxes obtained after object detection for tracking. The algorithm employs Kalman filtering to predict bounding boxes and then uses the Hungarian algorithm for matching between objects and trajectories. Its innovation lies in prioritizing the matching of high-confidence bounding boxes with existing trajectories and then matching low-confidence bounding boxes with unassociated trajectories. During this process, it restores true objects and filters out interference factors by comparing the similarity of low-confidence bounding boxes with previous trajectories. The algorithm's flowchart is depicted in Fig. 2 . Fig. 2 Basic flowchart of ByteTrack tracking algorithm. Fig. 2 | PMC467071_p20 | PMC467071 | The ByteTrack tracking algorithm | 1.494593 | other | Other | [
0.03728264570236206,
0.0005762516520917416,
0.9621410965919495
] | [
0.038922496140003204,
0.9592388272285461,
0.0011939627584069967,
0.0006446719053201377
] | en | 0.999998 |
The primary idea behind the ByteTrack tracking algorithm is to create tracking trajectories and use these trajectories to match objects in each frame, forming complete trajectories frame by frame. While the ByteTrack tracking algorithm demonstrates commendable performance in object tracking, there are several aspects that are worth optimizing and improving. Firstly, ByteTrack has not fully explored the potential of more superior object detectors. The quality of the object detector directly impacts the performance ceiling of the tracking algorithm. Secondly, ByteTrack does not utilize a feature re-identification network, which neglects the appearance information of the objects. In practical applications, this results in suboptimal tracking performance. Lastly, the ByteTrack tracking algorithm uses traditional linear Kalman filters with the same measurement noise scale for all objects, without considering the differences in the quality of different detection results. This makes it susceptible to the influence of low-quality detection results, potentially affecting the accuracy of state estimation (see Table 1 ). | PMC467071_p21 | PMC467071 | The ByteTrack tracking algorithm | 1.884732 | other | Other | [
0.04249047860503197,
0.0003913737600669265,
0.9571182131767273
] | [
0.31338149309158325,
0.6816304326057434,
0.0038971344474703074,
0.001090938807465136
] | en | 0.999996 |
The EBTrack algorithm has been specifically developed to effectively monitor electric bicycles in intricate traffic situations. It incorporates the YOLOv7 object detector, the NSA Kalman filter, and the ResNetEB feature extraction network, in addition to a customized matching mechanism. The comprehensive structure of the EBTrack algorithm is depicted in Table 2 as a pseudocode. By processing video sequences as its input, the algorithm generates the trajectories of electric bicycles, providing details, such as their bounding boxes, identifiers, and feature data. Table 1 Literature review summary: Pros and cons of various techniques for object tracking. Table 1 Method Advantages Disadvantages SiamFC End-to-end fully convolutional network for object position estimation Struggles with significant appearance changes and large search regions CFNet Combines correlation filtering with Siamese networks Sensitive to background clutter and occlusions SiamMa Integrates semantic segmentation for robust tracking Complex network structure; sensitive to scale changes DSST Efficient and robust algorithm Not suitable for real-time applications CCOT Accurate and robust tracking High computational complexity ECO Balances speed and accuracy Struggles with fast-moving objects ATOM Adaptive online learning for robust tracking Sensitive to occlusions and similar-looking objects Dimp Discriminative model prediction for accurate tracking High computational cost SORT Simple and fast algorithm Unstable object ID associations and frequent ID switches DeepSORT Cascade matching and feature re-identification network Compromises real-time performance due to feature extraction ByteTrack Effectively handles low-confidence bounding boxes Relies solely on bounding box information without considering appearance features MOTR Global multi-object tracking using Transformer architecture Computationally intensive and challenging for terminal deployment Zihan P et al. Recognizes wrong-way riding behavior of electric bicycles Limited to background subtraction and centroid analysis Xiaoping W et al. Enhances MDnet algorithm for complex traffic scenarios May struggle with fast-moving objects and significant appearance changes Zhenxiao L et al. Introduces LSTM motion model for vehicle tracking Does not address issues with frequent ID switching Caihong L et al. Provides cross-view visualization of tracked objects Complex implementation and limited to stitched field-of-view Table 2 Pseudo-code of EBTrack algorithm. Table 2 | PMC467071_p22 | PMC467071 | Method | 2.296759 | other | Study | [
0.02060823142528534,
0.0004665040469262749,
0.9789252281188965
] | [
0.6260291934013367,
0.28473609685897827,
0.08685288578271866,
0.0023818162735551596
] | en | 0.999998 |
To address the issues present in the ByteTrack tracking algorithm, the proposed EBTrack tracking algorithm starts by utilizing a high-performance object detector, YOLOv7, as the foundation for electric bicycle tracking. YOLOv7 is known for its high precision and lightweight design, enabling accurate detection of electric bicycle positions and providing reliable input data for subsequent tracking. Furthermore, the EBTrack tracking algorithm is built upon the ByteTrack tracking algorithm by introducing a feature extraction network tailored for re-identifying electric bicycles. | PMC467071_p23 | PMC467071 | The EBTrack tracking algorithm | 1.309881 | other | Other | [
0.020387867465615273,
0.0005050078616477549,
0.979107141494751
] | [
0.09961164742708206,
0.8983995318412781,
0.0009480165899731219,
0.001040783477947116
] | en | 0.999995 |
To ensure real-time performance, feature extraction is only performed when generating new object IDs, preventing frequent feature extraction from negatively impacting the tracking algorithm's real-time capabilities. Additionally, for more accurate trajectory predictions as well as enhanced tracking precision and stability, the algorithm draws inspiration from traditional linear Kalman filters and introduces an adaptive modulated noise scale Kalman filter . Given that, in real traffic scenarios, electric bicycles typically enter the monitoring frame from near or far rather than suddenly appearing in the center of the frame, the algorithm incorporates prior knowledge and designs a specialized matching mechanism to reduce the issue of ID switching caused by occlusions and other factors. | PMC467071_p24 | PMC467071 | The EBTrack tracking algorithm | 1.620828 | other | Other | [
0.046639181673526764,
0.0007148528238758445,
0.9526459574699402
] | [
0.05520638823509216,
0.9432592988014221,
0.0008627306669950485,
0.0006714814808219671
] | en | 0.999996 |
The EBTrack tracking algorithm is designed through the fusion of the YOLOv7 object detector, NSA Kalman filter, and the object re-identification network ResNetEB, along with a matching mechanism. This design enables the algorithm to demonstrate excellent performance in terms of accuracy and real-time tracking, particularly in complex scenarios involving object occlusion and appearance changes. The pseudocode for the EBTrack tracking algorithm is presented in Table 2 , which encompasses feature extraction and the matching mechanism. The algorithm takes, as input, a video sequence V , the YOLOv7 object detector, NSA Kalman filter, detection of confidence thresholds Thigh and Tlow, different regions within the frames Fmargin and Fmiddle, and the re-identification network ResNetEB. The algorithm's output is the object trajectory T , where each trajectory includes a bounding box, a sole ID, and feature information from the moment of object generation. | PMC467071_p25 | PMC467071 | EBTrack tracking algorithm basic workflow | 2.56927 | other | Other | [
0.14783409237861633,
0.0006461355951614678,
0.8515198230743408
] | [
0.41589516401290894,
0.5814580321311951,
0.0018646542448550463,
0.0007821757462807
] | en | 0.999997 |
The algorithm utilizes YOLOv7 as the object detector, performing object detection in each frame of the video. Firstly, the detection results are categorized into two classes based on the bounding box confidence, including high confidence and low confidence ( Table 2 , from line 3 to line 13). Secondly, the NSA Kalman filter is employed for position prediction and is updated ( Table 2 , from line 14 to line 16). During the initial association phase, high-confidence bounding boxes are used to perform IoU matching with all existing trajectories ( Table 2 , from line 17 to line 22). In the new trajectory generation phase, the object's position within the frame is taken into consideration. If the object is located at the frame's edge, feature information is directly extracted using the re-identification network ResNetEB, and a new trajectory is created. If the object is in the middle of the frame, ResNetEB is also used to extract feature information, and a third association is performed. This association step matches the feature information with all existing trajectories. If a successful match is made, the object is assigned to an existing trajectory. If a successful match cannot be made even after two frames, a new trajectory is created for the object ( Table 2 , from line 23 to line 38). | PMC467071_p26 | PMC467071 | EBTrack tracking algorithm basic workflow | 2.230165 | other | Other | [
0.08452658355236053,
0.0006509406375698745,
0.9148224592208862
] | [
0.251688688993454,
0.7458853125572205,
0.0015606241067871451,
0.000865375273860991
] | en | 0.999997 |
During the process of tracking electric bicycles, there are instances where objects are either occluded or not correctly detected by the object detector. This can result in objects briefly disappearing and then reappearing in the detector's field of view. Relying solely on the detector's results in such situations can lead to the issue of frequent object ID switching, where the same actual object may be assigned multiple IDs during the tracking process, significantly impacting subsequent behavior recognition. To address this issue, in the EBTrack algorithm, a re-identification network has been introduced. Its primary function is to extract the appearance information of objects, capturing their external characteristics to differentiate between the identities of different objects. This, in turn, helps overcome the problem of frequent object ID switching during the tracking process. | PMC467071_p27 | PMC467071 | The re-identification network ResNetEB | 1.21433 | other | Other | [
0.01019271370023489,
0.0004996811621822417,
0.9893075823783875
] | [
0.030396495014429092,
0.9681719541549683,
0.0007299674907699227,
0.0007016632007434964
] | en | 0.999998 |
By analyzing the original DeepSORT re-identification network, it was observed that it consists of 6 residual blocks, ultimately outputting features with a dimension of 128. However, this network structure is relatively simple, making it challenging to accurately capture real-time changes in the appearance of electric bicycles. As a result, there are noticeable limitations in the re-identification task for electric bicycle objects. | PMC467071_p28 | PMC467071 | The re-identification network ResNetEB | 1.252032 | other | Other | [
0.02847164124250412,
0.0004977824282832444,
0.971030592918396
] | [
0.08047740906476974,
0.9170977473258972,
0.0014991596108302474,
0.0009256380144506693
] | en | 0.999998 |
To address the aforementioned issue, two key improvement measures were implemented. Firstly, the feature dimension increased from 128 to 512 to enhance feature granularity and classification accuracy, thus strengthening the discrimination capability of the tracking algorithm. This enhancement allows the network to more accurately capture changes in the appearance of electric bicycles, contributing to improved re-identification performance. Secondly, to better accommodate the feature extraction requirements of electric bicycles in complex scenarios while maintaining real-time performance, the network's depth increased, named ResNetEB. The specific network structure is illustrated in Table 3 . This improvement aims to more effectively capture changes in the appearance of electric bicycles to meet the practical needs of tracking tasks. The combined effect of these two improvement measures enables the re-identification network to better meet the requirements of electric bicycle object tracking tasks while maintaining real-time performance. Table 3 ResNetEB network structure. Table 3 Layer Output Layer Output Convolutional 32*256*128 Residual 128*32*16 Convolutional 32*256*128 Down sampling residual 256*16*8 Max pooling 32*128*64 Residual 256*16*8 Residual 32*128*64 Down sampling residual 512*8*4 Down sampling residual 64*64*32 Residual 512*8*4 Residual 64*64*32 Fully connected 512 Down sampling residual 128*32*16 | PMC467071_p29 | PMC467071 | The re-identification network ResNetEB | 1.861963 | other | Study | [
0.031909920275211334,
0.00040477290167473257,
0.9676852822303772
] | [
0.6996011137962341,
0.2979471981525421,
0.0015047260094434023,
0.000946941610891372
] | en | 0.999997 |
The structure of a standard residual block in ResNetEB is illustrated in Fig. 3 (a) which comprises a main branch and a residual branch. In the figure, k represents the convolution's kernel size, s denotes the stride, and c indicates the number of channels. The main branch consists of two convolutional modules. It starts with a 3 × 3 convolutional kernel with a stride of 1, and the number of channels matches the input. Subsequently, BN (Batch Normalization) is applied, followed by activation function through the ReLU (Rectified Linear Unit) function . Next, the data passes through the second convolutional module and is added to the residual branch, followed by activation function through the ReLU function. The standard residual block does not alter the dimensions or the number of channels in the feature map. When down sampling is required, the structure of the down sampling residual block is shown in Fig. 3 (b). In the down sampling residual block, the first convolutional layer in the main branch employs a 3 × 3 convolutional kernel with a stride of 2, doubling the number of channels for down sampling. This operation reduces the feature map's size by half and doubles the number of channels. The residual branch utilizes a 1 × 1 convolutional kernel with a stride of 2, doubling the number of channels to decrease the size of the input feature map. Finally, the main branch and the residual branch are added together and activated using the ReLU function. Fig. 3 Residual block comparison of (a) Residual block and (b) Down sampling residual block. Fig. 3 | PMC467071_p30 | PMC467071 | The re-identification network ResNetEB | 3.594221 | other | Study | [
0.4592475891113281,
0.0007779847946949303,
0.5399744510650635
] | [
0.8392011523246765,
0.1584578901529312,
0.0019187440630048513,
0.0004221774870529771
] | en | 0.999998 |
The use of a residual structure is beneficial for maintaining the integrity of input information, reducing information loss in the forward propagation process, which is common in traditional convolutional layers. Furthermore, the network only needs to learn the differential parts between the input and output, simplifying the complexity of network training and improving convergence speed. In addition, the use of residual blocks effectively addresses the issues of gradient vanishing and exploding during network backpropagation. Finally, a fully connected layer is employed for classification training in the re-identification network. Once the training stage has been accomplished, the fully connected layer can be ignored, allowing for the direct matching of the feature vectors extracted by the network with the appearance information of the object and historical appearance information of trajectories. The advantages of this structure lie in its effectiveness and adaptability. | PMC467071_p31 | PMC467071 | The re-identification network ResNetEB | 3.364686 | other | Other | [
0.2618308365345001,
0.0009864801540970802,
0.7371827363967896
] | [
0.21720744669437408,
0.7771115899085999,
0.005136251449584961,
0.0005446805735118687
] | en | 0.999996 |
Therefore, by increasing the feature dimension to 512 in the re-identification network of the original DeepSORT, the algorithm has enhanced feature granularity and classification accuracy. Additionally, by increasing the network's depth and designing the ResNetEB network, it can better capture changes in the appearance of electric bicycles while maintaining real-time performance. This design adjustment allows the algorithm to better adapt to the task of tracking electric bicycles. | PMC467071_p32 | PMC467071 | The re-identification network ResNetEB | 1.340103 | other | Other | [
0.03280223533511162,
0.0006780146504752338,
0.9665197134017944
] | [
0.03489699587225914,
0.9636155962944031,
0.0008454090566374362,
0.0006419898709282279
] | en | 0.999997 |
In object tracking algorithms, the motion prediction module is a crucial component. Currently, tracking algorithms typically model the object's motion using a Kalman filter . However, linear Kalman filters have a limitation in that they use the same measurement noise scale for all objects, regardless of the quality of different detections. This results in low-quality detection results easily affecting the accuracy of state estimation. When the noise scale is larger, the detection weights in the state update phase become smaller, leading to increased uncertainty. To obtain more precise motion states and enhance the robustness of subsequent associations, a Kalman filter with adaptive noise scale adjustment has been employed , referred to as NSA KF. It can adaptively adjust the noise scale during the update process based on the confidence of the detection results to achieve more accurate motion state estimation. Specifically, the constant noise covariance is replaced with adaptively computed noise covariance. The formula is shown in Equation : (1) R k ∼ = ( 1 − c k ) R k In the prior formula, R k represents the measurement noise of predetermined constant covariance, and c k represents the detection confidence at state k . While the Kalman filter with adaptive noise scale adjustment is a relatively simple method, it significantly improves object tracking performance, particularly when dealing with various detection confidences. It effectively enhances the accuracy and robustness of object tracking. | PMC467071_p33 | PMC467071 | Kalman filter with adaptive modulation noise scale | 3.637068 | other | Study | [
0.22115951776504517,
0.0007385702920146286,
0.778101921081543
] | [
0.7126964330673218,
0.2737703025341034,
0.013001061975955963,
0.0005321731441654265
] | en | 0.999995 |
In complex traffic surveillance scenarios, electric bicycles typically enter the monitoring frame from a distant or near locations rather than suddenly appearing at the center of the frame. This prior knowledge serves as a crucial basis for the current research. To better meet application requirements, a specialized matching mechanism has been designed, as illustrated in Fig. 4 . Fig. 4 Special matching mechanism introducing prior knowledge. Fig. 4 | PMC467071_p34 | PMC467071 | Introducing a specialized matching mechanism with prior knowledge | 1.050554 | other | Other | [
0.011740255169570446,
0.0005302635836414993,
0.9877294898033142
] | [
0.028391990810632706,
0.9702304005622864,
0.0006573533173650503,
0.0007202071719802916
] | en | 0.999997 |
While ensuring that the electric bicycle detector meets the requirements for detection of accuracy and speed, the appearance of a new object ID in the monitoring frame typically indicates that the tracking algorithm may have lost a previously tracked object. In such cases, the following approach has been employed. | PMC467071_p35 | PMC467071 | Introducing a specialized matching mechanism with prior knowledge | 1.118175 | other | Other | [
0.013497023843228817,
0.0009475096594542265,
0.9855554103851318
] | [
0.0055540781468153,
0.99358069896698,
0.00043700874084606767,
0.00042818894144147635
] | en | 0.999999 |
Firstly, when a new object is about to appear at the center of the monitoring frame, the feature re-identification network has been used to extract the new object's feature information and match it with the feature information of recently existing objects. This process is crucial as it aids in more accurately identifying and tracking the object. The used features encompass appearance features and interactive features with the surrounding environment to ensure that the obtained information is highly distinctive and reliable. | PMC467071_p36 | PMC467071 | Introducing a specialized matching mechanism with prior knowledge | 1.287488 | other | Other | [
0.026155861094594002,
0.0011412393068894744,
0.9727028608322144
] | [
0.005613596644252539,
0.9936109185218811,
0.00044968404108658433,
0.0003257794014643878
] | en | 0.999998 |
Secondly, during the object matching process, several challenges are often encountered, including the presence of factors like occlusion, leading to temporary losses of objects during the tracking. Therefore, when a new object is about to appear in the central region of the monitoring frame, a delayed matching strategy is employed. The core idea of this strategy is that if a successful match with a previously lost object is not achieved even after two frames, the generation of a new object ID is allowed. This two-frame time window sufficiently accounts for the brief loss of an object in the frame, reducing the issue of ID switching caused by factors such as occlusion. The introduction of this strategy contributes to improving the stability and reliability of the tracking algorithm, reducing the generation of false object IDs and repeated object IDs. | PMC467071_p37 | PMC467071 | Introducing a specialized matching mechanism with prior knowledge | 1.57434 | other | Other | [
0.029441673308610916,
0.0007238866528496146,
0.9698343873023987
] | [
0.03457299992442131,
0.9641093015670776,
0.0007946986588649452,
0.0005230872193351388
] | en | 0.999996 |
Furthermore, to maintain the real-time performance of the tracking algorithm, the appearance features are extracted only when a new object is initially generated and during the delayed matching period for the soon-to-be generated object. This strategy maximizes the utilization of the feature re-identification network while ensuring the real-time capability of the tracking algorithm, minimizing the impact of ID switching issues. | PMC467071_p38 | PMC467071 | Introducing a specialized matching mechanism with prior knowledge | 1.410718 | other | Other | [
0.04702321067452431,
0.0010939935455098748,
0.9518827795982361
] | [
0.017001209780573845,
0.9820488095283508,
0.0004959508078172803,
0.00045397045323625207
] | en | 0.999997 |
The EBTrack tracking algorithm presents numerous benefits, such as the utilization of YOLOv7 for efficient and precise object detection, the enhancement of trajectory prediction through the NSA Kalman filter, and the reduction of ID switching via the specialized matching mechanism. In complex scenarios, the ResNetEB feature extraction network further improves the algorithm's performance by offering dependable feature representation for electric bicycles. | PMC467071_p39 | PMC467071 | Introducing a specialized matching mechanism with prior knowledge | 1.467295 | other | Other | [
0.05189475789666176,
0.0007296651601791382,
0.9473755359649658
] | [
0.018162716180086136,
0.9805077910423279,
0.0008782473742030561,
0.0004512058512773365
] | en | 0.999999 |
For algorithm training and validation, the hardware and software platform environment used consists of the following specifications: GPU: NVIDIA GeForce RTX 3060 Laptop GPU; CPU: 11th Gen Intel Core i5-11260H @ 2.60 GHz Hexa-core; VRAM: 6 GB; RAM: 32 GB; Operating System: Windows 10; Deep Learning Framework: PyTorch 1.12; and Programming Language: Python. | PMC467071_p40 | PMC467071 | Experimental results and analysis | 1.717995 | other | Other | [
0.3597971498966217,
0.001704737776890397,
0.6384981274604797
] | [
0.02342892996966839,
0.9757351875305176,
0.00044926724513061345,
0.00038666074397042394
] | en | 0.999998 |
Currently, there is no existing dataset for electric bicycle detection and tracking. To meet the requirements of training and validation of electric bicycle tracking algorithms, data have been collected and annotated from existing traffic intersection cameras under various conditions, including different locations, time periods, scenes, and weather conditions. This dataset encompasses several key features, including video data of electric bicycle objects during traffic peaks and off-peak hours, daytime and nighttime, and rainy weather conditions. The total duration of the videos amounts to 500 h, with a resolution of 1920 × 1080, and frames were captured at a rate of 15 frames per second. Additionally, 10,000 original images have been extracted, the electric bicycle objects from these video data. After data cleaning and augmentation, two distinct datasets were created, namely one for electric bicycle object detection (Dataset-Det) and another for electric bicycle re-identification (Dataset-ID). Finally, 30 segments of diverse and representative video data have been randomly selected from the original videos, each lasting 5 min. This subset was used as the validation dataset for tracking algorithms (Dataset-Track). Fig. 5 displays some sample images from the electric bicycle dataset. Fig. 5 Partial images of electric bicycle dataset. Fig. 5 | PMC467071_p41 | PMC467071 | Dataset | 1.294565 | other | Study | [
0.022701947018504143,
0.0005320552736520767,
0.9767659902572632
] | [
0.9402474164962769,
0.05812534689903259,
0.0009087725775316358,
0.0007184530841186643
] | en | 0.999996 |
In this study, the following evaluation metrics were used to assess algorithm performance, namely MOTA (Multiple Object Tracking Accuracy), IDF1 (ID F1-Score), FPS (Frames Per Second), Precision, and Recall. | PMC467071_p42 | PMC467071 | Evaluation metrics | 2.131239 | other | Study | [
0.41130536794662476,
0.0014590610517188907,
0.5872355699539185
] | [
0.9720762968063354,
0.026643164455890656,
0.0008457436924800277,
0.0004349203663878143
] | en | 0.999997 |
MOTA (Multiple Object Tracking Accuracy) is one of the most widely used evaluation metrics in for object tracking, aimed at providing a comprehensive assessment of the performance of object tracking algorithms. MOTA takes into account three primary sources of tracking errors: FP (False Positives), FN (False Negatives), and ID (Identity) Switches, all of which are crucial in practical object tracking scenarios. Specifically, the formula for MOTA is as shown in Equation . (2) M O T A = 1 − ∑ t ( F P t + F N t + I D s t ) ∑ t G T t where, F P t represents the number of false positive detections in the t t h frame, indicating instances where the algorithm incorrectly marks the background or non-object as objects. F N t represents the number of false negative detections in the t t h frame, signifying instances where actual objects exist but are not detected by the algorithm. I D s refers to the number of identity switch errors in the t t h frame, which occur when an object's identity changes during tracking due to mismatches or occlusions. G T t stands for the number of true objects in the t t h frame. MOTA primarily focuses on evaluating the performance of the object detector. When the detector performs well and there are fewer ID switch errors, the MOTA value will be higher. | PMC467071_p43 | PMC467071 | MOTA (Multiple Object Tracking Accuracy) | 2.85272 | other | Other | [
0.08861066401004791,
0.0005409151781350374,
0.9108483195304871
] | [
0.3790643811225891,
0.6136648058891296,
0.006518148351460695,
0.0007525837863795459
] | en | 0.999996 |
The IDF 1 metric places more emphasis on association. It is a comprehensive evaluation metric for object tracking that focuses on both IDP (ID Precision) and IDR (ID Recall). It particularly emphasizes on the continuity of tracking and the accuracy of identity information. It measures the extent to which the tracking algorithm can maintain both continuity and accuracy throughout the entire tracking period. A higher IDF1 value indicates higher accuracy in tracking specific objects, meaning that the algorithm can consistently and accurately track the same object across different frames. It primarily assesses whether the tracking algorithm can maintain the initially created trajectories continuously and consistently. The IDF1 formula is shown in Equation . (3) I D F 1 = 2 | I D T P | 2 | I D T P | + | I D F P | + | I D F N | where, IDTP (True Positive ID) represents the number of times the tracking algorithm correctly maintains the identity information of the object across adjacent frames. It indicates instances where the algorithm correctly identifies the same object's identity information across different frames. IDFP (False Positive ID) represents the number of times the tracking algorithm erroneously associates different objects with the same identity information. In other words, it quantifies how often the algorithm mistakenly confuses the identity information of different objects. IDFN (False Negative ID) represents the number of times the tracking algorithm fails to correctly maintain the identity information of the object. It indicates instances where the algorithm incorrectly loses track of the object's identity information. | PMC467071_p44 | PMC467071 | IDF 1(Identification F1) | 3.043911 | other | Other | [
0.11493269354104996,
0.0007032843423075974,
0.8843640089035034
] | [
0.29307880997657776,
0.701302170753479,
0.004993609618395567,
0.0006254615145735443
] | en | 0.999997 |
FPS is used to measure the real-time performance and efficiency of object tracking algorithms. In academic research and practical applications, FPS is typically used to assess the speed of object tracking algorithms when processing real-time video streams. | PMC467071_p45 | PMC467071 | FPS（Frames per second） | 1.319964 | other | Other | [
0.047798220068216324,
0.0009753096383064985,
0.9512264132499695
] | [
0.0044395350851118565,
0.9943006038665771,
0.0009533671545796096,
0.0003064630145672709
] | en | 0.999998 |
Precision refers to the ratio of the number of correctly detected objects by the detector to the total number of detections. It signifies the accuracy of the detector, indicating how many of the detection results are correct, as shown in Equation . (4) P r e c i s i o n = T P T P + F P here, TP (True Positives) represents the number of correctly detected objects, and FP (False Positives) represents the number of incorrectly detected objects. | PMC467071_p46 | PMC467071 | Precision | 3.167797 | biomedical | Other | [
0.6656105518341064,
0.0009878459386527538,
0.33340156078338623
] | [
0.24610480666160583,
0.75139319896698,
0.002082768129184842,
0.00041916937334463
] | en | 1 |
Recall refers to the ratio of the number of correctly detected objects by the detector to the total number of true objects. It indicates how many of the true objects the detector is able to find, as shown in Equation . (5) R e c a l l = T P T P + F N | PMC467071_p47 | PMC467071 | Recall | 1.517012 | other | Other | [
0.1467178612947464,
0.0014652329264208674,
0.8518169522285461
] | [
0.05339789390563965,
0.9445372223854065,
0.0012829047627747059,
0.0007819760940037668
] | en | 0.999998 |
Such that, FN (False Negatives) represents the number of targets that were not detected by the detector. | PMC467071_p48 | PMC467071 | Recall | 2.754801 | biomedical | Study | [
0.9853798747062683,
0.000690440705511719,
0.013929593376815319
] | [
0.49961793422698975,
0.4979294240474701,
0.0015816140221431851,
0.0008710096590220928
] | en | 0.999997 |
To validate the performance and characteristics of the proposed EBTrack tracking algorithm, an analysis is conducted from several perspectives. Firstly, a comparative evaluation of different object detectors is performed to determine the detector that performs optimally in electric bicycle tracking. Secondly, discussions are carried out regarding re-identification networks with different feature dimensions to analyze their impact on tracking performance. Thirdly, module ablation experiments are conducted to analyze the roles and importance of individual modules within the algorithm. Lastly, the EBTrack algorithm is compared with various classical tracking algorithms. | PMC467071_p49 | PMC467071 | Experimental results and analysis | 1.66802 | other | Study | [
0.036100663244724274,
0.0006781240808777511,
0.9632211327552795
] | [
0.9410654902458191,
0.05657259374856949,
0.0015525026246905327,
0.000809345452580601
] | en | 0.999996 |
In object tracking, the performance and accuracy of tracking algorithms are influenced by the quality of the object detector and the size of the dataset. In this study, a self-constructed dataset called Dataset-Det was used to train the object detector. During the training process, different detectors were iterated for 200 times to ensure the creation of high-quality detectors, providing reliable input for subsequent tracking algorithms. The ByteTrack tracking algorithm was introduced to the trained object detectors, and a performance comparison was conducted on the Dataset-Track validation set, as shown in Table 4 . Precision and Recall represent the results of different object detectors on validation set of the Dataset-Det. When high-precision object detectors are selected, the performance of the tracking algorithm also improves. Therefore, this study chose to use the lightweight YOLOv7 as the detector for the EBTrack tracking algorithm to ensure its tracking performance reaches the optimal state in practical applications. Table 4 Results of using different detectors in dataset-track validation set. Table 4 Detector MOTA/% IDF1/% IDs FPS/s Precision/% Recall/% SSD 62.9 65.4 291 34.5 79.8 77.6 F-RCNN 67.4 70.3 275 23.1 83.4 81.5 YOLOv3 72.3 75.1 243 36.7 88.7 86.6 YOLOv5 78.1 81.3 222 37.6 91.1 90.4 YOLOX-Tiny 74.5 77.6 241 40.9 89.7 88.1 YOLOX 79.8 82.1 213 37.0 92.0 90.5 YOLOv7 84.4 88.6 190 38.6 96.9 96.1 | PMC467071_p50 | PMC467071 | Comparison of different object detectors | 3.441724 | other | Study | [
0.33086198568344116,
0.0008715151925571263,
0.6682665348052979
] | [
0.9935703277587891,
0.005741512402892113,
0.0005381236551329494,
0.00015002561849541962
] | en | 0.999996 |
Balancing accuracy and real-time performance is a key concern. In this study, improvements were made to the existing DeepSORT re-identification network branch, and iterations of 200 times were performed on different feature dimensions using the Dataset-ID dataset. Subsequently, these enhancements were integrated into the EBTrack tracking algorithm, and performance comparisons were conducted on the Dataset-Track validation set. Precision and Recall represent the results of different feature dimensions of the re-identification network on the Dataset-ID validation set. As shown in Table 5 , using 1024-dimensional re-identification feature dimensions resulted in relatively higher IDF1 and MOTA scores, indicating that as the re-identification feature dimension increases, the network's feature extraction capability also improves. Higher-dimensional features can often better represent the appearance characteristics of targets, thereby enhancing tracking accuracy. Considering the real-time requirements of the electric bicycle tracking algorithm, setting the re-identification feature dimension to 512 effectively improves the network's feature extraction capability while maintaining real-time performance. This balance takes into account the performance required in complex tracking tasks and ensures the feasibility and practicality of the electric bicycle tracking algorithm. Table 5 Results of using different feature dimensions for re-recognition networks in the Dataset-Track validation set. Table 5 Dimension MOTA/% IDF1/% IDs FPS/s Precision/% Recall/% 1024 90.4 95.1 158 32.0 98.6 97.8 512 89.8 94.2 164 34.1 98.1 97.2 256 85.1 89.0 180 34.8 95.2 93.9 128 82.0 85.6 186 35.4 91.4 90.3 64 80.2 82.4 193 35.9 85.9 83.1 | PMC467071_p51 | PMC467071 | Comparison of Re-identification networks with different feature dimensions | 3.884767 | biomedical | Study | [
0.5995044112205505,
0.0010666640009731054,
0.39942899346351624
] | [
0.9985087513923645,
0.0010454814182594419,
0.00037132741999812424,
0.00007441412162734196
] | en | 0.999995 |
To validate the effectiveness of the major modules in the EBTrack tracking algorithm, the performance of different modules has been compared on validation set of the Dataset-Track. As shown in Table 6 , after introducing the YOLOv7 object detector, a significant improvement was observed in MOTA and IDF1, and a decreasing trend was observed in ID switches. This performance enhancement can be attributed to the strong dependence of the tracking algorithm on the results of the object detector. Therefore, an excellent object detector is crucial for the overall performance improvement of the tracking algorithm. Experimental results also showed an improvement in detection performance after introducing the NSA Kalman filter. With the addition of the ResNetEB feature re-identification network and the use of a specific matching mechanism, MOTA reached 89.8 %, and IDF1 reached 94.2 %, while significantly reducing the frequency of ID switches. Although the feature re-identification network consumes some computational resources, leading to a slight decrease in FPS that the reduction is within an acceptable range. Table 6 Results of module ablation experiment. Table 6 YOLOv7 NSA KF ResNetEB MOTA/% IDF1/% IDs FPS/s × × × 79.8 82.1 213 37.0 ✓ × × 84.4 88.6 190 38.6 ✓ ✓ × 84.6 89.0 187 38.5 ✓ ✓ ✓ 89.8 94.2 164 34.1 | PMC467071_p52 | PMC467071 | Module ablation experiments | 3.618265 | biomedical | Study | [
0.5328369736671448,
0.0009556087316013873,
0.46620747447013855
] | [
0.9969285130500793,
0.002643631072714925,
0.0003196877078153193,
0.00010817128350026906
] | en | 0.999997 |
In this study, comparative experiments were conducted with the SORT, DeepSORT, FairMOT, and ByteTrack algorithms on validation set of the Dataset-Track, and the experimental results are presented in Table 7 . The SORT tracking algorithm exhibits frequent ID switches, which often prevent the continuous tracking of complete electric bicycle trajectories, resulting in a large number of fragmented trajectories that significantly interfere with subsequent behavior recognition accuracy. The DeepSORT tracking algorithm, while introducing a feature re-identification network and employing a cascade matching approach, has relatively improved the issues found in the SORT algorithm. However, the feature re-identification network demands significant computational resources, leading to a severe decrease in real-time performance. Both ByteTrack and FairMOT tracking algorithms show performance improvements compared to the previous two algorithms, but ByteTrack only uses bounding box information for tracking and does not utilize target appearance information. FairMOT combines detection and tracking within the same model, which results in slower terminal running speeds. The proposed EBTrack tracking algorithm, with the presence of the feature re-identification network, has a certain impact on real-time performance. However, it significantly reduces the frequency of ID switches and lessens trajectory fragmentation, providing a better data foundation for subsequent behavior recognition. Table 7 Results of different tracking algorithms on validation set of the Dataset-Track. Table 7 Method MOTA/% IDF1/% IDs FPS/s SORT 70.1 70.3 293 39.4 DeepSORT 73.0 77.1 246 21.7 FairMOT 78.1 80.4 219 25.8 ByteTrack 79.8 82.1 213 37.0 EBTrack 89.8 94.2 164 34.1 | PMC467071_p53 | PMC467071 | Comparison of different tracking algorithms | 3.679947 | other | Study | [
0.4204539656639099,
0.0012712341267615557,
0.5782747268676758
] | [
0.9979792237281799,
0.00145382818300277,
0.0004643380525521934,
0.00010259317059535533
] | en | 0.999997 |
Table 7 represents the results of the EBTrack tracking algorithm on validation set of the Dataset-Track. It's worth noting that the issue of ID switches is relatively prominent in densely crowded and nighttime environments. | PMC467071_p54 | PMC467071 | Comparison of different tracking algorithms | 1.202919 | other | Other | [
0.040140409022569656,
0.0006172987050376832,
0.9592423439025879
] | [
0.2022148072719574,
0.7945472598075867,
0.0019349146168679,
0.001303026219829917
] | en | 0.999998 |
In the following, here are the extended experimental results. - Comparison with Additional Tracking Algorithms: | PMC467071_p55 | PMC467071 | Comparison of different tracking algorithms | 1.554418 | biomedical | Other | [
0.6285833716392517,
0.002734244568273425,
0.3686824142932892
] | [
0.4069480895996094,
0.585529625415802,
0.005497677717357874,
0.0020246400963515043
] | en | 0.999999 |
Tracktor : This algorithm utilizes a tracking-by-detection approach and incorporates a novel tracking-based branch to predict the bounding box location in the next frame. On the Dataset-Track validation set, Tracktor achieved an MOTA of 75.6 % and an IDF1 of 78.3 %. | PMC467071_p56 | PMC467071 | Comparison of different tracking algorithms | 1.280592 | other | Other | [
0.02497422695159912,
0.0007420760812237859,
0.9742836952209473
] | [
0.051462605595588684,
0.946492612361908,
0.0011054158676415682,
0.0009393333457410336
] | en | 0.999997 |
JDE : This is a joint detection and embedding method that combines detection and re-identification into a single network. JDE obtained an MOTA of 78.9 % and an IDF1 of 82.4 % on the Dataset-Track validation set. | PMC467071_p57 | PMC467071 | Comparison of different tracking algorithms | 1.653289 | other | Other | [
0.2046235352754593,
0.0010789919178932905,
0.794297456741333
] | [
0.10092858970165253,
0.8966426253318787,
0.0015595940640196204,
0.0008692152914591134
] | en | 0.999995 |
TransTrack : This algorithm employs a Transformer-based architecture for multi-object tracking. TransTrack achieved an MOTA of 81.2 % and an IDF1 of 84.6 % on the Dataset-Track validation set. - Comparison with Different Feature Dimensions: | PMC467071_p58 | PMC467071 | Comparison of different tracking algorithms | 1.333338 | other | Other | [
0.059882137924432755,
0.0008716742740944028,
0.9392462372779846
] | [
0.08636848628520966,
0.910901665687561,
0.0016490640118718147,
0.0010807339567691088
] | en | 0.999997 |
By decreasing the feature dimension to 256, the EBTrack algorithm was able to achieve an MOTA of 85.1 % and an IDF1 of 89.0 %. This reduction in feature dimensionality led to a minor decline in tracking accuracy but enhanced real-time performance . | PMC467071_p59 | PMC467071 | Comparison of different tracking algorithms | 2.092212 | other | Study | [
0.4969852864742279,
0.001735151745378971,
0.5012795329093933
] | [
0.795592188835144,
0.20193926990032196,
0.0013700993731617928,
0.0010983935790136456
] | en | 0.999996 |
On the other hand, by increasing the feature dimension to 1024, the EBTrack algorithm achieved an MOTA of 90.4 % and an IDF1 of 95.1 %. Although this higher feature dimension improved tracking accuracy, it also resulted in higher computational complexity. - Comparison with Different Object Detectors: | PMC467071_p60 | PMC467071 | Comparison of different tracking algorithms | 1.803794 | other | Study | [
0.2760748565196991,
0.0014184946194291115,
0.7225066423416138
] | [
0.500745415687561,
0.49545201659202576,
0.0022493842989206314,
0.0015532433753833175
] | en | 0.999999 |
YOLOv5 : Using YOLOv5 as the object detector, the EBTrack algorithm achieved an MOTA of 78.1 % and an IDF1 of 81.3 % on the Dataset-Track validation set. | PMC467071_p61 | PMC467071 | Comparison of different tracking algorithms | 1.687275 | other | Other | [
0.17043426632881165,
0.0011634292313829064,
0.8284022808074951
] | [
0.2995965778827667,
0.6975849270820618,
0.001596168614923954,
0.0012223167577758431
] | en | 0.999998 |
Faster R–CNN : With Faster R–CNN as the object detector, EBTrack obtained an MOTA of 67.4 % and an IDF1 of 70.3 %, demonstrating the impact of detector performance on tracking accuracy. - Ablation Study: | PMC467071_p62 | PMC467071 | Comparison of different tracking algorithms | 1.97113 | biomedical | Study | [
0.6388823986053467,
0.0013632309855893254,
0.35975441336631775
] | [
0.8299380540847778,
0.16803991794586182,
0.0011965460143983364,
0.0008254993008449674
] | en | 0.999996 |
EBTrack without NSA Kalman Filter: The removal of the NSA Kalman filter resulted in EBTrack achieving an MOTA of 84.4 % and an IDF1 of 88.6 %. The absence of the NSA Kalman filter led to slightly lower accuracy in trajectory predictions. | PMC467071_p63 | PMC467071 | Comparison of different tracking algorithms | 2.259219 | biomedical | Study | [
0.7027774453163147,
0.0012593354331329465,
0.29596322774887085
] | [
0.8098811507225037,
0.18809252977371216,
0.001243394915945828,
0.0007829547394067049
] | en | 0.999996 |
EBTrack without ResNetEB: Excluding the ResNetEB feature extraction network led to EBTrack achieving an MOTA of 84.6 % and an IDF1 of 89.0 %. The absence of the ResNetEB network impacted the algorithm's performance in handling complex scenarios involving occlusions and appearance changes. | PMC467071_p64 | PMC467071 | Comparison of different tracking algorithms | 1.671591 | other | Other | [
0.21135428547859192,
0.001433506840839982,
0.7872121334075928
] | [
0.28122103214263916,
0.7155680060386658,
0.0017540563130751252,
0.0014568407787010074
] | en | 0.999994 |
These comparative analyses highlight the robustness and efficiency of the EBTrack tracking algorithm in electric bicycle tracking applications. | PMC467071_p65 | PMC467071 | Comparison of different tracking algorithms | 1.437195 | other | Other | [
0.05987745150923729,
0.000887296861037612,
0.9392351508140564
] | [
0.3148331940174103,
0.678462564945221,
0.00509493425488472,
0.0016093595186248422
] | en | 0.999998 |
In this study, an effective algorithm was introduced for tracking electric bicycles, named EBTrack, which is specifically tailored for traffic monitoring situations. The algorithm used the lightweight YOLOv7 as the object detector, ensuring precise and dependable object detection. The incorporation of the ResNetEB feature extraction network enhanced the algorithm's performance in intricate scenarios characterized by high pedestrian density and occlusion. The adaptive modulated noise scale Kalman filter boosted the accuracy and stability of object trajectories, enabling the algorithm to adjust to dynamic environments. Furthermore, the revamped association matching mechanism successfully reduced the problem of object ID switching, thereby enhancing tracking stability and continuity. The experimental findings validated the efficacy of EBTrack, achieving an MOTA of 89.8 % and an IDF1 of 94.2 %. This algorithm laid a solid groundwork for subsequent behavior recognition tasks in traffic monitoring scenarios. Nevertheless, it is crucial to recognize the constraints of the current study. Firstly, EBTrack is primarily custom-made for tracking electric bicycles and may necessitate further modifications for other vehicle types or scenarios. Secondly, the algorithm presupposes the availability of high-quality video data, and its performance may deteriorate under low-light or challenging weather conditions. Lastly, the real-time performance of EBTrack hinges on the computational resources, necessitating potential optimizations for deployment on resource-constrained devices. In future endeavors, it is intended to address these limitations and expand the capabilities of the EBTrack algorithm to encompass a broader spectrum of traffic monitoring applications, thereby enhancing its adaptability and robustness. Additionally, advanced machine learning techniques and sensor fusion will be explored to further refine tracking accuracy and adaptability. Moreover, the plan can include providing the incorporation of efficient machine learning methodologies, such as reinforcement learning and graph neural networks, to enhance the precision and resilience of the EBTrack algorithm. | PMC467071_p66 | PMC467071 | Conclusion | 3.362666 | other | Study | [
0.14315330982208252,
0.0009195468737743795,
0.8559271097183228
] | [
0.9945026636123657,
0.0041864169761538506,
0.0011196095729246736,
0.00019132389570586383
] | en | 0.999998 |
This research was funded by Natural Science Research Project of Anhui Province ，Natural Science Research Project of 10.13039/501100012404 Fuyang Normal University | PMC467071_p67 | PMC467071 | Funding | 0.993294 | other | Other | [
0.013377070426940918,
0.0011998236877843738,
0.9854230880737305
] | [
0.0015212633879855275,
0.9975581169128418,
0.0005257150041870773,
0.0003948990779463202
] | en | 0.999996 |
All data generated or analysed during this study are included in this published article. | PMC467071_p68 | PMC467071 | Data availability | 0.785522 | biomedical | Other | [
0.8804402947425842,
0.0018458575941622257,
0.1177138015627861
] | [
0.04161475971341133,
0.9562581181526184,
0.0011576457181945443,
0.0009695389890111983
] | en | 0.999998 |
Zhengyan Liu: Investigation. Chaoyue Dai: Investigation. Xu Li: Investigation. | PMC467071_p69 | PMC467071 | CRediT authorship contribution statement | 0.880195 | other | Other | [
0.15350206196308136,
0.0036497183609753847,
0.8428481221199036
] | [
0.005479262210428715,
0.9932492971420288,
0.0006758665549568832,
0.0005955773522146046
] | en | 0.999995 |
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. | PMC467071_p70 | PMC467071 | Declaration of competing interest | 0.981821 | other | Other | [
0.004888109862804413,
0.0006584162474609911,
0.994453489780426
] | [
0.001968854106962681,
0.9968124032020569,
0.0006366003071889281,
0.0005821675877086818
] | en | 0.999997 |
Funding for this research was supported by NIH grant RO1-HL62150 to AHL. In addition, this work was supported by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Chemical Sciences, Geosciences and Biosciences Division, under award #DE-SC0015662. | 39022598_p0 | 39022598 | Funding information | 0.980187 | other | Other | [
0.009711837396025658,
0.001354293548502028,
0.9889339208602905
] | [
0.0010016950545832515,
0.9982014894485474,
0.0004285174945835024,
0.0003682735550682992
] | en | 0.999998 |
The authors declare they have no known competing financial interests or personal relationships that may affect this work. | 39022598_p1 | 39022598 | Declaration of competing interest | 0.999675 | other | Other | [
0.004005456808954477,
0.0007528123678639531,
0.9952417612075806
] | [
0.002134499838575721,
0.9964466691017151,
0.0007240998093038797,
0.0006948150112293661
] | en | 0.999996 |
Theodore J. Kottom: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Eva M. Carmona: Writing – review & editing, Writing – original draft. Bernd Lepenies: Resources. Andrew H. Limper: Writing – review & editing, Writing – original draft, Supervision, Funding acquisition. | 39022598_p2 | 39022598 | CRediT authorship contribution statement | 0.951842 | other | Other | [
0.16470760107040405,
0.0036922066938132048,
0.8316002488136292
] | [
0.004090885166078806,
0.9950554370880127,
0.0004124108236283064,
0.0004412748967297375
] | en | 0.999997 |
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Andrew H Limper reports financial support was provided by National Institutes of Health. Theodore J Kottom reports financial support was provided by National Institutes of Health. Eva Carmona reports financial support was provided by National Institutes of Health. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. | 39022598_p3 | 39022598 | Declaration of competing interest | 0.998679 | other | Other | [
0.004610390402376652,
0.0007837062003090978,
0.9946058392524719
] | [
0.0013227553572505713,
0.9975948929786682,
0.0005705606308765709,
0.0005116990068927407
] | en | 0.999997 |
Microphthalmia is defined as a small, underdeveloped eye caused by disrupted eye development through genetic or environmental factors in the first trimester. Clinical phenotypic heterogeneity exists in patients with varying severity and associated ocular and systemic features. As one of the most severe developmental eye abnormalities, microphthalmia accounts for approximately 3% to 12% of blinding cases in children, 1 with the prevalence ranging from 10.0 to 10.8 per 10,000 births. 2 There is no treatment for microphthalmia that will restore vision. | 39007834_p0 | 39007834 | Introduction | 3.878931 | biomedical | Review | [
0.9982660412788391,
0.001223152969032526,
0.0005108735640533268
] | [
0.04447270184755325,
0.1811796873807907,
0.7713629007339478,
0.0029846711549907923
] | en | 0.999997 |
Besides a small proportion of cases that are attributed to environmental factors, such as intrauterine infections and toxins, genetic alterations are the major causes of such a disease. 3 Genes implicated in main non-syndromic microphthalmia include SOX2 , OTX2 , RAX , VSX2 , STRA6 , RARB , ALDH1A3 , MAB21L2 , VAX1 , BMP7 , GDF3 , and GDF6 . 3 , 4 More genes up to approximately 100 are associated with systemic microphthalmia. Due to the conserved ocular development and physiology between mice and humans, dozens of mouse lines with a microphthalmia trait have been generated, most of which resulted from the disruption of the genes identified in human patients. These genes included Sox2 , Otx2 , Rax , Vsx2 , Pax6 , Stra6 , Foxe3 , Bmp4 , Bmp7 , Smoc1 , Shh , Porcn , Foxc1 , Fras1 , Frem1 , Tctn2 , Col4a1 , Tbc1d32 , Prss56 , Pxdn , Pitx2 , Pitx3 , Mitf , Cryaa , Frem2 , Rpgrip1l , Smg9 , Snx3 , Dag1 , Hmx1 , Rere , and Rab18 (reviewed from Mouse Genome Informatics database: http://www.informatics.jax.org/ ). The mouse offers the possibility to genetically test the roles of modifiers and single nucleotide polymorphisms (SNPs); these aspects open new avenues for ophthalmogenetics in the mouse. Overall, 237 causative genes in mice and 98 in human subjects share 31 overlapping genes, as illustrated in Figure 1 . The complete gene list is provided in Supplementary Table S1 . | 39007834_p1 | 39007834 | Introduction | 4.278555 | biomedical | Study | [
0.9992783665657043,
0.00044974748743698,
0.00027192794368602335
] | [
0.8186802268028259,
0.0014024372212588787,
0.17939074337482452,
0.0005265141371637583
] | en | 0.999997 |
Various studies suggest that Hedgehog ( HH ) signaling plays essential roles in human and mouse eye development. 5 Mutations/deletions in human Sonic Hedgehog ( SHH ) cause holoprosencephaly, including anophthalmia, cyclopia, and coloboma in severe cases. 6 – 8 Homozygous Shh null mutant mice show that Shh plays a critical role in the brain and spinal cord, the axial skeleton, and the limbs, and SHH was required to separate the eye field into bilateral domains. 9 Other components of the HH signaling pathway, such as PTCH1 and CDON, have been reported to cause microphthalmia when mutated in human patients. Such evidence suggests that HH signaling is closely related to eye development and the cause of microphthalmia. 10 , 11 Interestingly, constitutively active HH signaling in surface ectoderm of mice caused by a mutation of Smo , the coding protein of which is repressed by PTCH1 when not activated, results in aberrant and disorganized lens and retina morphology. 12 | 39007834_p2 | 39007834 | Introduction | 4.534348 | biomedical | Study | [
0.9992856383323669,
0.0003925147757399827,
0.0003218383062630892
] | [
0.938631534576416,
0.00078806426608935,
0.060187067836523056,
0.0003932523541152477
] | en | 0.999996 |
Transcriptional factors GLI1, GLI2, and GLI3 are thought to regulate most of the transcriptional responses to HH signaling. 13 Different from GLI1, which acts predominantly as positive regulators of target genes in HH signaling, GLI2 and GLI3 play either an activating or a repressing role depending on the HH signal availability. 13 – 17 GLI2 plays a stronger activating role than GLI3 in the HH signaling cascade. 16 , 18 In contrast, the repressing part of GLI3 is more predominant than that of GLI2. 18 – 20 Regarding GLI3, in the absence of any HH signal, the C-terminal region of GLI3 is cleaved after amino acid 700 to generate an N-terminal 83 kDa transcriptional repressor (GLI3-R). 21 In the presence of HH, GLI3 is in the full-length form that functions as a transcriptional activator (GLI3-A). 22 Because GLI3 primarily acts as a transcriptional repressor, the loss of GLI3 is often functionally equated to the overactivity of the HH pathway. 23 In the in vivo studies of GLI3 functioning in eye development, 2 transgenic mouse lines have been used: Gli3 +/Xt-J and Gli3 ∆699/∆699 . In Xt-J (Extra-toes J ) homozygotes, Gli3 expression is completely missing during embryogenesis, and the mice would die within 2 days after birth. 24 , 25 In the heterozygous Gli3 +/Xt-J line, the embryos exhibit eye defects varying from microphthalmia to anophthalmia with significantly smaller lenses or no lenses at all. 26 , 27 In the Gli3 ∆699/∆699 line, the repressor form of GLI3 (GLI3-R) is constitutively expressed. 28 Although the Gli3 ∆699/∆699 mice die shortly after birth, 28 by contrast with the ocular phenotype in the embryos of Gli3 Xt-J/Xt-J mice, Gli3 ∆699/∆699 embryos do not exhibit any morphological defects in the eye, 18 indicating that GLI3-R but not GLI3-A is essential for eye development. | 39007834_p3 | 39007834 | Introduction | 4.664035 | biomedical | Study | [
0.9992181062698364,
0.0004143148835282773,
0.00036755777546204627
] | [
0.9952665567398071,
0.0005435955245047808,
0.004007916431874037,
0.00018194742733612657
] | en | 0.999996 |
It is still an open question how GLI3 regulates eye formation, it would be of great importance to investigate the dosage-dependent function of GLI3 in eye development. Our group have long been interested in the lens development and lens-related diseases, in this context, we generated a lens-specific TgGli3Ki/Ki mouse line. In this line, chicken βB1-crystallin promoter (−434/+30) 29 – 31 is conjugated with full-length cDNA of Gli3 , endowing the active expression of GLI3 in lens fiber cells. We found that the homozygous TgGli3Ki/Ki mice are viable for at least 12 months (to the end of the observation point) with severe microphthalmia, and they are all blind due to total synechia of the iris. These results suggest that overexpression of GLI3 in the lens disrupts eye development, and the size of the whole eye is deeply affected by the lens size. | 39007834_p4 | 39007834 | Introduction | 4.16224 | biomedical | Study | [
0.9996356964111328,
0.0001883602380985394,
0.00017587824549991637
] | [
0.9992063641548157,
0.0003734546189662069,
0.00033741723746061325,
0.00008280693145934492
] | en | 0.999996 |
The first mouse microphthalmia transcription factor ( Mitf ) mutation was discovered over 60 years ago, which was originated from a cohort of irradiated mice. 32 Since then, most mouse models identified with a microphthalmia phenotype were created by forward genetics. For example, in a spontaneous mouse mutant line Pitx3 416insG , 33 the mutant mice have closed eyelids with no apparent eyes (anophthalmia) or very small eyes (microphthalmia). Recent progress in targeted genome editing makes it much easier to directly modify specific genes. In a transgenic Pax6 mouse line, where the downstream regulatory region of Pax6 was disrupted, it presented a similar eye development pattern of microphthalmia and aniridia. 34 Another study generated a gene-dosage allelic series of Sox2 mutations in the mouse, suggested that a reduction of SOX2 expression to <40% of normal causes variable microphthalmia. 35 Although microphthalmia trait manifest, at least partially, in all these models, the severity of the phenotypes are highly variable, from isolated mild microphthalmia to anophthalmia. The homogeneous TgGli3Ki/Ki mouse line in this study provides a consistent tool to explore the pathology of microphthalmia without impacting other organs of the body. | 39007834_p5 | 39007834 | Introduction | 4.400205 | biomedical | Study | [
0.9995610117912292,
0.00023821633658371866,
0.00020081418915651739
] | [
0.9973909258842468,
0.0003282986581325531,
0.00216927332803607,
0.00011149503552587703
] | en | 0.999996 |
The animal studies were conducted in accordance with the ARVO Animal Statement. We used C57BL/6JGpt mice as the background line. We made TgGli3Ki/Ki knockin mice via CRISPR/Cas9 system. The mouse Gli3 gene was inserted into Hipp 11 (H11) locus via Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)/CRISPR-associated 9 (Cas9) system. The detailed description of the construction of this line is provided in the Results section. | 39007834_p6 | 39007834 | Generation of the TgGli3Ki/Ki Mice | 3.88637 | biomedical | Study | [
0.9995294809341431,
0.0001356184802716598,
0.0003349177713971585
] | [
0.9930035471916199,
0.006446849089115858,
0.0004053461307194084,
0.00014423498942051083
] | en | 0.999997 |
Total RNA was extracted using TRIzol and measured on NanoDrop 2000 (Thermo Fisher Scientific). Reverse transcription was performed using an RT kit , and quantitative polymerase chain reaction (qPCR) was performed using SYBR Premix . The qPCR was performed on the ABI 7500 PCR machine, and data were analyzed using the ABI 7500 software version 2.0.6 (Life Technologies, Thermo Fisher Scientific). Actb was used as the endogenous control gene. The 2 −∆∆ct method was applied for the relative quantification. The primer sequences are provided in Supplementary Table S2 . The acronyms for all the genes are provided in Supplementary Table S3 . | 39007834_p7 | 39007834 | Quantitative Polymerase Chain Reaction | 4.088737 | biomedical | Study | [
0.9996535778045654,
0.00016174513439182192,
0.00018471521616447717
] | [
0.9983299374580383,
0.0012827699538320303,
0.00031023425981402397,
0.00007709229248575866
] | en | 0.999996 |
Protein extracts from eye tissues were subjected to SDS-PAGE and blotted onto 0.45 µm PVDF membrane (Merck Millipore), incubated overnight with the primary antibody GLI3 at 1 in 200 dilution, or β-Actin, at 1 in 20,000 dilution in Tris-buffered saline with 0.1% Tween-20 and 5% milk. The anti-goat second antibody was used to visualize the protein bands. The membrane was developed using the Chemilum HRP Substrate . | 39007834_p8 | 39007834 | Western Blot | 4.073973 | biomedical | Study | [
0.9995272159576416,
0.00022794681717641652,
0.0002448097802698612
] | [
0.9902738928794861,
0.008984764106571674,
0.0005631595849990845,
0.00017814170860219747
] | en | 0.999998 |
For hematoxylin and eosin (H&E) staining of mice tissues, dissected eyeballs were fixed in 10% formalin overnight, followed by dehydration through an ethanol gradient. Tissues were embedded in paraffin and sectioned at 5 µm. To observe the morphology of the entire eyeball, the tissue sections were stained with H&E. To deparaffinize the tissues, paraffin sections were incubated for 1 hour at 60°C in xylene twice for 10 minutes, washed in 100% ethanol twice for 3 minutes, followed by incubation in 95%, 80%, and 70% ethanol for 3 minutes in each step. To retrieve the antigens, the slides were incubated in 0.125% trypsin at 37°C for 30 minutes. Then, the slides can be used to perform immunofluorescent staining the same as the frozen sections. | 39007834_p9 | 39007834 | Histology and Immunofluorescence | 4.070139 | biomedical | Study | [
0.9989946484565735,
0.0007094983011484146,
0.0002958571130875498
] | [
0.7966310977935791,
0.19995816051959991,
0.00224325736053288,
0.001167455455288291
] | en | 0.999996 |
Subsets and Splits