xxxmy

@zhengdq99 您可以使用 https://github.com/ChingHo97/FCOS-PyTorch-37.2AP 这个仓库的代码，他fix了我的一些错误

@zhengdq99 不可以哦，直接resize，会导致物体形状完全变了，可能把一个东西拉的很长或很扁，跟实际测试相差很大

I got a bad test result after training. I've figured out that there's a problem of preprocessing. So I have fixed this problem, you can try to train by yourself....

The eval code is from network. There seems to be something wrong with this. Thanks!

Based on the description in the paper. It's implemented according to my own understanding while it's similar to the original implementation.

don't need remove it, I fix the bug then it can work well.

High mAP in papers always due to many tricks and data augmentation. My implementation doesn't use any data augmentation, which is very important for getting a good result. And batch...

It's really not easy to train a network. The difference between 0.67 and 0.69 is really small, and they are both not good result on VOCdataset. Did you train it...

A common approach is to train on VOC07 and VOC12 train+val, then eval on VOC07 test. The dataset setting you used is rare. 0.02 mAP gap can not measure good...

It depends on whether you train it from scratch and how large the batch size is. You can set epochs up a little bigger,such as 50 or 100, and then...