experiencor / keras-yolo3

Training and Detecting Objects with YOLO3
MIT License
1.6k stars 861 forks source link

Question about mAP calculation #315

Closed brian123214 closed 2 years ago

brian123214 commented 3 years ago

In utils.utils, the function evaluate looks like this:

def evaluate(model, generator, iou_threshold=0.5, obj_thresh=0.5, nms_thresh=0.45, net_h=416, net_w=416, save_path=None) My questions is, is this calculating AP at IOU at 0.5 or AP as how it is calculated in the COCO dataset, which is slightly different? https://jonathan-hui.medium.com/map-mean-average-precision-for-object-detection-45c121a31173 How would you implement the other AP calculation that is not used in this repo?