Tianshui Chen

We have presented the detailed structure of the encoding and decoding layers in our paper. Please read [our paper](https://arxiv.org/abs/1712.07493) for more details.

We do not use BN after convolutional layers of the WAE. We merely use one BN layer when feeding the decomposed sub-image into the classification network. We train the WAE...

Yes, we have tried resizing the input image into two quarter-resolution sub-images, but the results are not satisfactory. We are now exploring these.

Yes, we directly use their results in the paper for comparison.

> Veg-fru dataset has a test set which is much bigger than the train set and val set. I'd like to know if you can tell me which two of...

We copy the results of HybridNet and CB-CNN from table 5 in the VegFru paper. In the caption, the authors clearly state that the results are evaluated on the test...

At this repository, we merely release the codes and models for testing. We will release the training codes soon.