Update README.md
Browse files
README.md
CHANGED
@@ -46,4 +46,83 @@ configs:
|
|
46 |
path: data/train-*
|
47 |
---
|
48 |
|
49 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
path: data/train-*
|
47 |
---
|
48 |
|
49 |
+
## Load data
|
50 |
+
|
51 |
+
```python
|
52 |
+
import datasets
|
53 |
+
|
54 |
+
dataset = datasets.load_dataset("mrdbourke/trashify_manual_labelled_images")
|
55 |
+
dataset
|
56 |
+
```
|
57 |
+
|
58 |
+
## View a sample
|
59 |
+
|
60 |
+
```python
|
61 |
+
dataset["train"][0]
|
62 |
+
```
|
63 |
+
|
64 |
+
Output:
|
65 |
+
|
66 |
+
```
|
67 |
+
{'image': <PIL.Image.Image image mode=RGB size=960x1280>,
|
68 |
+
'image_id': 292,
|
69 |
+
'annotations': {'file_name': ['00347467-13f1-4cb9-94aa-4e4369457e0c.jpeg',
|
70 |
+
'00347467-13f1-4cb9-94aa-4e4369457e0c.jpeg'],
|
71 |
+
'image_id': [292, 292],
|
72 |
+
'category_id': [1, 0],
|
73 |
+
'bbox': [[523.7000122070312,
|
74 |
+
545.0999755859375,
|
75 |
+
402.79998779296875,
|
76 |
+
336.1000061035156],
|
77 |
+
[10.399999618530273,
|
78 |
+
163.6999969482422,
|
79 |
+
943.4000244140625,
|
80 |
+
1101.9000244140625]],
|
81 |
+
'iscrowd': [0, 0],
|
82 |
+
'area': [135381.078125, 1039532.4375]},
|
83 |
+
'label_source': 'manual_prodigy_label',
|
84 |
+
'image_source': 'manual_taken_photo'}
|
85 |
+
```
|
86 |
+
|
87 |
+
**Note:** Boxes in "bbox" key are in `XYWH` format or `[x_min, y_min, box_width, box_height]`. If you'd like them in `XYXY` format, you'll have to convert them.
|
88 |
+
|
89 |
+
## Get categories
|
90 |
+
|
91 |
+
```python
|
92 |
+
# Get the categories from the dataset
|
93 |
+
# Note: this requires the dataset to have been uploaded with this feature setup
|
94 |
+
categories = dataset["train"].features["annotations"].feature["category_id"]
|
95 |
+
|
96 |
+
# Get the names attribute
|
97 |
+
categories.names
|
98 |
+
|
99 |
+
>>> ['bin', 'hand', 'not_bin', 'not_hand', 'not_trash', 'trash', 'trash_arm']
|
100 |
+
```
|
101 |
+
|
102 |
+
## Create label2id and id2label
|
103 |
+
|
104 |
+
```python
|
105 |
+
id2label = {i: class_name for i, class_name in enumerate(categories.names)}
|
106 |
+
label2id = {value: key for key, value in id2label.items()}
|
107 |
+
|
108 |
+
id2label, label2id
|
109 |
+
```
|
110 |
+
|
111 |
+
Output:
|
112 |
+
|
113 |
+
```
|
114 |
+
({0: 'bin',
|
115 |
+
1: 'hand',
|
116 |
+
2: 'not_bin',
|
117 |
+
3: 'not_hand',
|
118 |
+
4: 'not_trash',
|
119 |
+
5: 'trash',
|
120 |
+
6: 'trash_arm'},
|
121 |
+
{'bin': 0,
|
122 |
+
'hand': 1,
|
123 |
+
'not_bin': 2,
|
124 |
+
'not_hand': 3,
|
125 |
+
'not_trash': 4,
|
126 |
+
'trash': 5,
|
127 |
+
'trash_arm': 6})
|
128 |
+
```
|