Includes evaluation.py

[PT-10]

• Evaluates model predictions using IoU-based matching and multiple AP thresholds.
• Calculates key detection metrics: Precision, Recall, F1 Score, [email protected], [email protected], [email protected]
• Outputs results to both console logs and a evaluation_results.txt file for reproducibility.

Additional Dependencies:

• numpy, but not added in requirements.txt since installing torch should cover it.

[sergiopaniego]

Thanks for the PR!
Since we received previously a PR for the same task, maybe we can coordinate about this task in that PR. Could you both coordinate in order to contribute?

[PT-10]

Thanks for the PR! Since we received previously a PR for the same task, maybe we can coordinate about this task in that PR. Could you both coordinate in order to contribute?

I believe the PR has gone dormant, I can take it over, also do you want me to include the baseline evaluation metrics of the gemma model loaded in a quantized form or metrics for the smol-vlm model