BobLiu20 / YOLOv3_PyTorch

Full implementation of YOLOv3 in PyTorch
624 stars 191 forks source link

mAP problem #23

Open Yanxingang opened 6 years ago

Yanxingang commented 6 years ago

I test the mAP result: 1: official_yolov3_weights_pytorch: (AP,IoU=0.5): 0.540 2: yolov3_weights_pytorch: (AP,IoU=0.5): 0.469, which is lower than what you says in 'weights/readme':0.5966

bobo0810 commented 6 years ago

hello,I also have this question. Can you solve it?

bobo0810 commented 6 years ago

@Yanxingang python eval.py params.py instead of python eval_coco.py params.py , you will get the expected results. This is all based on the premise that the mAP code is correct。

jayden199012 commented 6 years ago

The mAP calculation seems quite different from any voc or coco implementation

ajoseph12 commented 5 years ago

I believe the mAP code is flawed. The code seems to be calculating an average recall of sorts. Nevertheless a great repo :)