essentials_for_CIL

essentials_for_CIL copied to clipboard

I expect to get the result of CCIL+H-Aug 71.66, but I get 68.25. I conduct the following command ``` python main_cifar.py \ --new-classes 10 \ --start-classes 50 \ --cosine \...

wusuoweima

I am currently using 128 gigabytes of RAM, but out of memory comes out when I train with Imagenet. How many gigabytes of ram did you experiment with?

th0214

I can't find where is "coarse_label_names" used in the code?

cuiheng1234

Hi Sudhanshu, Thank you for your nice work and codebase. If possible, can you please share the version of ImageNet 1000 that you have used? Is it ImageNet 2017? If...

JosephKJ

About icarl_construct_exemplar_set function

1

Hi, I think the following codes seem not proper: https://github.com/sud0301/essentials_for_CIL/blob/c7eb378c822792977d96daa5dd98c0202263ec44/main_cifar.py#L321-L337 After debugging I find that when features[idx, :] = 0.0 implemented, exemplar_features is also set to 0, thus S is...

superduduguan

In 'main_imagenet.py' , line 321: x = Variable(transform(Image.fromarray(img))).cuda() You use Image.fromarray(img) to load a array type image object. However, channel order in img is BGR (opencv) while the channel order...

haoweiz23

I want to use the checkpoint.pth , load it and and want to do predition , can you please share a small snippet for this

ahsan856jalal