Running test_RFB.py with your provided model, RFBNet300_VOC_80_7.pth, get a mAP of 0.8056 not 80.7% posted in your github.

GOATmessi8 / RFBNet

Receptive Field Block Net for Accurate and Fast Object Detection, ECCV 2018

MIT License

1.41k stars 357 forks source link

Open ZhiqiJiang opened 5 years ago

ZhiqiJiang commented 5 years ago

The command is python3 test_RFB.py --trained_model RFBNet300_VOC_80_7.pth. And the result is showed as below: Screenshot from 2019-04-12 10-55-46

love1314pds commented 5 years ago

hellow,I run the command same with yours, but my APs of 20 classes and mAP are 0. Did you meet the same problem before the current results.

ZhiqiJiang commented 5 years ago

@ love1314pds No I didn't. Maybe I modified something correspondingly.

ZhiqiJiang commented 5 years ago

@love1314pds 我已经不止一次遇到跨框架引起精度损失这个问题了,就算是利用作者提供的pretrained model.我一直还没有耐下性子去定位分析乃至解决这个问题.

love1314pds commented 5 years ago

嗯，好的，谢谢

Feobi1999 commented 4 years ago

302f50c86b0ff30e845e6ca641bfe08 I use the same commad but ge't a map=0.459 But I didnot change anything~

KrystalCWT commented 4 years ago

Ps o

I have the same problem. Did you solve it?