Tsung-Wei Ke

Yes, according to our empirical experience, it is necessary to exponentially decrease the affinity field loss during training.

Hi, please check our [bashscripts examples](https://github.com/twke18/Adaptive_Affinity_Fields/blob/master/bashscripts/voc12/train_pspnet_aaf.sh). The parameters listed in the bash files are the ones we reported in the paper. For training, perhaps you can first try training the...

The script assumes that you have per-pixel semantic segmentation labels. For inference under weakly-supervised settings, there are two steps: 1) using the learned features to propagate softmax classifier predictions ([this...

Thanks for the suggestion! The requirements.txt is uploaded correspondingly.

Great suggestion! Since `pip freeze > requirements.txt` lists many packages which are not used for the project, we instead create another file `our_requirements.txt` for listing our complete python environment to...

In our experience, we do not observe performance gap among different versions of PyTorch. Which experiments are you currently trying to reproduce? Do you have exactly the same number of...

If you are using a GPU card, have you tried to set GPUS=0 [here](https://github.com/twke18/HSG/blob/8de138f7d90ce4f87f29ca274e67f60e48ff09e2/bashscripts/coco/train.sh#L21)? Also, have you checked the occupancy of GPU memory? It's also likely that OOM occurred and...

@RobinLiuZX , do you still have the problem? @Lan-MC , are you also using a single GPU for training?