Pretrained Model and codes about dctnet

[kaix90]

I have checked all the links to the pretrained models, and all of them are accessible. Please try again.

You can find the pre-processing codes at https://github.com/calmevtime/DCTNet/blob/4c756ddf664512c73236619d06df3891f7489de3/classification/datasets/dataset_imagenet_dct.py#L108.

In our experiment, we leverage a heap map to visualize the averaged probability of each channel being selected. We also tested another heuristic selection method, which has similar performance as the learning-based selection method. Thanks.

[RainbowLLL]

Hello, I am trying to visualize the feature maps of for DCT input and RGB input to figure out the differences. So I'm wondering if you could share the pretrained DCT-192 model before channel sampling. Thanks in advance!

[SarfarazHabib]

Hi @RainbowLLL
I am also trying to visualize some feature maps of these DCT networks.Were you able to do it successfully. Can you help me with it ?

[lld533]

I have checked all the links to the pretrained models, and all of them are accessible. Please try again.

You can find the pre-processing codes at

https://github.com/calmevtime/DCTNet/blob/4c756ddf664512c73236619d06df3891f7489de3/classification/datasets/dataset_imagenet_dct.py#L108

.
In our experiment, we leverage a heap map to visualize the averaged probability of each channel being selected. We also tested another heuristic selection method, which has similar performance as the learning-based selection method. Thanks.

Hello,

I check the Google drive link to the pretrained models. I'm afraid that they are now stored in the recycle bin and NOT ready for download. Could you please recover it? Many thanks in advance!