-
I added CBAM and BAM in ResNet50 on fine grained dataset, e.g. CUB200. But it had a worse performance than pure ResNet50 backbone. Was it because I failed to have right training hyper parameters? Its …
-
The models in
"Or you can download the trained checkpoints [cub200_checkpoint.pth.tar] / [cars196_checkpoint.pth.tar]"
have a 128 embedded size, where is it possible to download the models for …
-
Hi,
I think line 12 in your code (in awa2.m, awa1.m, sun.m, cub200.m) should be deleted,
because before the transpose classAttributes is already num_classes-by-num_attributes.
data downloaded fr…
-
What is the difference between training with the two configurations (cub200_promptccd_w_gmp.yaml cub200_promptccd_w_gmp_unknown_C.yaml)
-
Dear developers,
Hello, I would like to ask if you can open the download path of the trained model used in your paper, thank you very much!
-
Thank you for sharing your code. Nice work! I have a question about the Binomial Deviance Loss. In your previous paper, Binomial without mining achieves recall-top1 of 64% on CUB200, but I can not rep…
-
When uploading a dataset to the clearml-server, the uploading statistics, in terms of file size uploaded, is wrong compared to the total size. The total file size appears to be correct, in this exampl…
-
First, when training cub training set, the code always stops automatically every 13epoch, for example, 0-13, 14-27. I only changed the batchsize to 4, and at the same time, I changed the learning rate…
-
你好,我想咨询一下两个问题,
一、cub数据集加载时候,box标签是不是并没有用到(因为用到算强监督)
二、pretrained_model加载是是resnet50,那么所初始化加载的resnet50(original)是否需要cub200上 fine-tune
-
You have done a very nice job on your paper! I tried to implement your proposed network these days, but I found several problems.
- The first one is I found that the Fc layer after descriptor alw…