isJc

请问你numpy用的什么版本，我使用requirements所要求的1.15.1会报错 

> > 请问你numpy用的什么版本，我使用requirements所要求的1.15.1会报错  > > 我用的是numpy 1.19.5 好的，谢谢

exactly! According to my own understanding, I divide the custom data set into base and novel samples, but the loss will not decrease after training. I don't know why! 

For mmfewshot, it's strange why there is no discussion

> > > if i set pretrained = 'open-mmlab://detectron2/resnet50_caffe' in configs/detection/base/models/faster_rcnn_r50_caffe_c4.py, the log print "unexpected key in source state_dict: conv1.weight, ......" > > > > > > More details are...

> What config did you use? `configs/detection/base/models/faster_rcnn_r50_caffe_c4.py` is not the whole config file. I haven't met this error in MMDet before thanks for your reply! after i reinstall the env...

For the doubts about ‘grid-0.5’, I see many such problems, eg #6252, #471... Compared with the previous code(y[..., 0:2] *2 - 0.5 + grid), I found that the step of...

> @isJunCheng grid computation now embeds offsets (after #7262) to reduce FLOPs in detect.py and simplify export models. The change has no mathematical implications, the result is exactly the same...