Datasets:
update_readme2
Browse files
README.md
CHANGED
|
@@ -31,4 +31,46 @@ language:
|
|
| 31 |
- th
|
| 32 |
size_categories:
|
| 33 |
- 1K<n<10K
|
| 34 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
- th
|
| 32 |
size_categories:
|
| 33 |
- 1K<n<10K
|
| 34 |
+
---
|
| 35 |
+
|
| 36 |
+
# ThaiOCRBench: A Task-Diverse Benchmark for Vision-Language Understanding in Thai
|
| 37 |
+
|
| 38 |
+
**ThaiOCRBench** is the first comprehensive benchmark for evaluating vision-language models (VLMs) on Thai text-rich visual understanding tasks.
|
| 39 |
+
Inspired by OCRBench v2, it contains **2,808 human-annotated samples** across **13 diverse tasks**, including table parsing, chart understanding, full-page OCR, key information extraction, and visual question answering.
|
| 40 |
+
|
| 41 |
+
The benchmark enables **standardized zero-shot evaluation** for both proprietary and open-source models, revealing significant performance gaps and paving the way for document understanding in low-resource languages.
|
| 42 |
+
|
| 43 |
+
🚀 **Our paper _ThaiOCRBench_ has been accepted to the IJCNLP-AACL 2025 Main Conference!**
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
## 📊 Dataset Statistics
|
| 47 |
+
|
| 48 |
+
| Task Type | Number of Samples |
|
| 49 |
+
|-------------------------------|-------------------|
|
| 50 |
+
| Text Recognition | 333 |
|
| 51 |
+
| Table Parsing | 193 |
|
| 52 |
+
| Full-page OCR | 197 |
|
| 53 |
+
| Chart Parsing | 200 |
|
| 54 |
+
| Key Information Extraction | 201 |
|
| 55 |
+
| Diagram VQA | 204 |
|
| 56 |
+
| Fine-grained Text Recognition | 206 |
|
| 57 |
+
| Handwritten Content Extraction| 209 |
|
| 58 |
+
| Key Information Mapping | 209 |
|
| 59 |
+
| Document Parsing | 211 |
|
| 60 |
+
| Infographics VQA | 213 |
|
| 61 |
+
| Document Classification | 215 |
|
| 62 |
+
| Cognition VQA | 217 |
|
| 63 |
+
| **Total** | **2,808** |
|
| 64 |
+
|
| 65 |
+
|
| 66 |
+
## 🧠 Performance of VLMs on ThaiOCRBench
|
| 67 |
+
|
| 68 |
+
<p align="center">
|
| 69 |
+
<img src="https://github.com/scb-10x/ThaiOCRBench/blob/main/pics/thaiocrbench_eval.png" width="88%" height="60%">
|
| 70 |
+
</p>
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
## 📘 Citation
|
| 74 |
+
|
| 75 |
+
If you use this benchmark in your research, please cite:
|
| 76 |
+
|