Datasets:

Modalities:
Image
Text
Formats:
parquet
Languages:
Arabic
ArXiv:
Libraries:
Datasets
pandas
License:
Omartificial-Intelligence-Space commited on
Commit
4191d32
Β·
verified Β·
1 Parent(s): f8d9004

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +47 -6
README.md CHANGED
@@ -33,15 +33,15 @@ size_categories:
33
 
34
  ## πŸ“‹ Dataset Summary
35
 
36
- The QARI Markdown Mixed Dataset is a specialized synthetic dataset designed for training Arabic OCR models with a focus on complex document layouts and HTML structure understanding.
37
  This dataset is part of the QARI-OCR project, which achieves state-of-the-art performance in Arabic text recognition.
38
 
39
- This dataset contains **50,000 synthetically generated Arabic document images** with corresponding ground truth text in HTML/Markdown format, featuring:
40
  - πŸ”€ **Full diacritical marks (tashkeel)** support
41
  - πŸ“ **Mixed font sizes** within documents (headers, body text, annotations)
42
  - 🎨 **12 distinct Arabic fonts** ranging from common Naskh to ornate calligraphic styles
43
  - πŸ“„ **Realistic document layouts** with structural HTML tags
44
- - πŸ–ΌοΈ **Three degradation levels**: Clean, Moderate, and Heavy
45
 
46
  ## 🎯 Intended Use
47
 
@@ -55,16 +55,57 @@ This dataset is specifically designed for:
55
 
56
  | Metric | Value |
57
  |--------|-------|
58
- | **Total Images** | 50,000 |
59
- | **Text Sources** | Modern news articles + Classical Islamic corpus |
 
 
 
60
  | **Font Variety** | 12 Arabic fonts |
61
  | **Font Size Range** | 14px - 100px |
62
- | **Degradation Types** | 3 (Clean, Moderate, Heavy) |
63
  | **Diacritics Support** | βœ… Full tashkeel |
64
  | **HTML Structure** | βœ… Preserved |
65
  | **Layout Complexity** | βœ… High (mixed sizes, headers) |
66
 
 
67
 
 
 
 
 
 
 
 
 
 
 
68
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
69
 
70
 
 
33
 
34
  ## πŸ“‹ Dataset Summary
35
 
36
+ The QARI v0.3 Markdown Mixed Dataset is a specialized synthetic dataset designed for training Arabic OCR models with a focus on complex document layouts and HTML structure understanding.
37
  This dataset is part of the QARI-OCR project, which achieves state-of-the-art performance in Arabic text recognition.
38
 
39
+ This dataset contains **37,000 synthetically generated Arabic document images** (29.6k train, 3.7k validation, 3.7k test) with corresponding ground truth text in HTML/Markdown format, featuring:
40
  - πŸ”€ **Full diacritical marks (tashkeel)** support
41
  - πŸ“ **Mixed font sizes** within documents (headers, body text, annotations)
42
  - 🎨 **12 distinct Arabic fonts** ranging from common Naskh to ornate calligraphic styles
43
  - πŸ“„ **Realistic document layouts** with structural HTML tags
44
+ - πŸ–ΌοΈ **Multiple text sources** including Basma2423 and YoussefAnwar Arabic news
45
 
46
  ## 🎯 Intended Use
47
 
 
55
 
56
  | Metric | Value |
57
  |--------|-------|
58
+ | **Total Images** | 37,000 |
59
+ | **Train Set** | 29,600 (80%) |
60
+ | **Validation Set** | 3,700 (10%) |
61
+ | **Test Set** | 3,700 (10%) |
62
+ | **Text Sources** | oddadmix/Basma2423-Text-with-Diacritics-Correction + YoussefAnwar/Arabic-news |
63
  | **Font Variety** | 12 Arabic fonts |
64
  | **Font Size Range** | 14px - 100px |
 
65
  | **Diacritics Support** | βœ… Full tashkeel |
66
  | **HTML Structure** | βœ… Preserved |
67
  | **Layout Complexity** | βœ… High (mixed sizes, headers) |
68
 
69
+ ## πŸ”§ Data Generation Pipeline
70
 
71
+ <div align="center">
72
+
73
+ | Stage | Process | Details |
74
+ |-------|---------|---------|
75
+ | **1. Text Collection** | Source gathering | Basma2423 (with diacritics) + YoussefAnwar Arabic news |
76
+ | **2. HTML Templating** | Layout generation | Mixed font sizes, structural elements |
77
+ | **3. Rendering** | WeasyPrint β†’ PDF β†’ Image | High-quality document rendering |
78
+ | **4. Degradation** | Synthetic noise | Clean / Moderate / Heavy variants |
79
+
80
+ </div>
81
 
82
+ ## πŸ“ˆ Model Performance
83
+
84
+ When used to train QARI v0.3, this dataset enables:
85
+
86
+ | Metric | Score |
87
+ |--------|-------|
88
+ | **Character Error Rate (CER)** | 0.300 |
89
+ | **Word Error Rate (WER)** | 0.485 |
90
+ | **BLEU Score** | 0.545 |
91
+ | **Training Time** | 11 hours |
92
+ | **COβ‚‚ Emissions** | 1.88 kg eq. |
93
+
94
+ ### Key Advantages:
95
+ - πŸ“ **Superior layout understanding** compared to plain text models
96
+ - 🏷️ **HTML tag preservation** for structured document conversion
97
+ - ⚑ **Resource efficient** - 5x less training time than larger datasets
98
+ - 🎯 **Specialized performance** for document structure tasks
99
+
100
+ ## Citation
101
+
102
+ ```markdown
103
+ @article{wasfy2025qari,
104
+ title={QARI-OCR: High-Fidelity Arabic Text Recognition through Multimodal Large Language Model Adaptation},
105
+ author={Wasfy, Ahmed and Nacar, Omer and Elkhateb, Abdelakreem and Reda, Mahmoud and Elshehy, Omar and Ammar, Adel and Boulila, Wadii},
106
+ journal={arXiv preprint arXiv:2506.02295},
107
+ year={2025}
108
+ }
109
+ ```
110
 
111