rafaelpadilla / review_object_detection_metrics

Object Detection Metrics. 14 object detection metrics: mean Average Precision (mAP), Average Recall (AR), Spatio-Temporal Tube Average Precision (STT-AP). This project supports different bounding box formats as in COCO, PASCAL, Imagenet, etc.
Other
1.08k stars 215 forks source link

possible incorrect equation #106

Closed karl-gardner closed 2 years ago

karl-gardner commented 2 years ago

Hello Rafael,

I'm wondering if the equation for 11 point interpolation could be wrong. If you think about the sum in the AP formula, wouldn't it be divided by 10 not 11 so wouldn't the correct term before the sum be: 1/(11-1). Because the width of every step in the 11 points is only 0.1 not 0.09 as you currently have. This would affect the final average precision value:

image

If you divide by 10 instead of 11 you would get 0.975 instead of 0.8864. Let me know if this makes sense or if I am wrong on this.

Thanks,

Karl Gardner

rafaelpadilla commented 2 years ago

Closing this issue as it is being discussed here