Ying
Ying
Hi, I do not know what the fmap is? fmap means image or image-feature?
关于类PA的作用
你好 很感谢你的工作。有一个问题想请教下:代码中class PA ()的作用是什么?感觉与论文2.4 Position Encoding不对应。因为代码中PA调用是在class PatchEmbed (认为与论文2.3 Patch Embedding相关)class BasicStem (the first patch embedding module)。综上 class PA ()与论文2.3 Patch Embedding相关,不与2.4 Position Encoding相关,但是论文2.4 Position Encoding中公式(8)描述了PA的Conv2d()和Sigmoid() 。 再次谢谢
python main.py --mode train --data_path --stage 2 RuntimeError: Given input size: (256x11x3). Calculated output size: (256x0x0). Output size is too small Can you tell me how to fix it. thanks...
Thanks for your excellent work. I am confusing about the four trained checkpoints(nl_center_param_epochs, nl_model_, nl_optimizer_,nl_optimizer_center) in GPS Folder, which is the GPS_market1501.pth in python test.py --config_file configs/market1501_gps_softmax_triplet_center.yml MODEL.PRETRAIN_CHOICE "('self')" TEST.WEIGHT...
What is the version of the environments? Would it be convenient to provide it? And the dataset?