CAMEL-Bench: Model Performance Across Vision Understanding Tasks
This table shows the performance of different models across various tasks including OCR, chart understanding, video, medical imaging, and more.
CAMEL-Bench Model Performance
Method | Average Score | MM Understanding & Reasoning | OCR & Document Understanding | Video Understanding | Remote Sensing Understanding | Charts & Diagram Understanding | Agro Specific | Cultural Specific Understanding | Medical Imaging |
---|---|---|---|---|---|---|---|---|---|
LLaVa-OneVision-7B | 63.77 | 56.78 | 72.35 | 64.09 | 45.92 | 62.35 | 85.05 | 78.09 | 43.77 |
Submission Instructions
To contribute your model's results to the CAMEL-Bench leaderboard:
Via GitHub Pull Request:
- Use this evaluation script to test your model and generate results.
- Create a pull request in the CAMEL-Bench GitHub repository with your results.
Via Email:
- Send your results to ahmed.heakl@mbzuai.ac.ae, and weโll add them to the leaderboard for you.
We look forward to seeing your contributions!