Dear Yichao,
On top of our recommendations below, that the paper should be either trained or evaluated on ImageNet, we would also like to request at least a bit of dedicated discussion about ImageNet in your submission. This could be, for example, a discussion on the importance of ImageNet as an evaluation benchmark for your method. Ideally, we would like the accepted papers be aligned well with the overall theme of the workshop.
Please don’t hesitate to get back with further questions.
Thank you!
Joon