search

YangFengSEU / CEDR

Contrastive Embedding Distribution Refinement and Entropy-Aware Attention for 3D Point Cloud Classification
4 stars 1 forks source link

Experiments issues #1

Open NeilCui6 opened 2 years ago

NeilCui6 commented 2 years ago

Dear @YangFengSEU ，

Many thanks for sharing the codes of your work. Could you please kindly let me know where can I find other py files for running the Experiments? There is only train.py available at the moment.

Thanks a lot.

YangFengSEU commented 2 years ago

Thank you for your interest. We will be releasing additional py code soon after the paper is accepted. For now only the test code and the model.t7 is available.

NeilCui6 commented 2 years ago

Many thanks for your reply, @YangFengSEU. Sorry, to make sure I understand your meaning correctly. Could I confirm something:

  1. Is the "test code" that you mentioned the train.py file?
  2. what's the model.t7?
  3. At this stage, with the current available resources here, am I able to run through a test successfully (for example with using pointnet as the backbone network)?

It seems you haven't uploaded some files like "contrastloss ", "util", right?

Thank you. : )

YangFengSEU commented 2 years ago

Thank you for asking.

  1. Yes, you can use the train.py for test. It contained test_loader in train.py.
  2. You can find the best model.t7 in https://drive.google.com/file/d/1nEAOU9hujcUW6I2mXjTxGoWlQqZfAja6/view?usp=sharing.
  3. Without the contrastloss, you can also run the test experiment with a backbone network. For our approach has no constraints on the network. But the mode.t7 we released is trained on pointnet++ and GBNet. When we release all the code, you can train your own model. You are welcome to ask any other questions you may have.
NeilCui6 commented 2 years ago

Thank you for asking.

  1. Yes, you can use the train.py for test. It contained test_loader in train.py.
  2. You can find the best model.t7 in https://drive.google.com/file/d/1nEAOU9hujcUW6I2mXjTxGoWlQqZfAja6/view?usp=sharing.
  3. Without the contrastloss, you can also run the test experiment with a backbone network. For our approach has no constraints on the network. But the mode.t7 we released is trained on pointnet++ and GBNet. When we release all the code, you can train your own model. You are welcome to ask any other questions you may have.

Many Thanks, @YangFengSEU. I have sent a request for accessing to the model.t7 in google drive. Could you please approve it?

YangFengSEU commented 2 years ago

Yes，it's accessible now.

zrcvae commented 1 year ago

Hello, can you share the follow-up model.py code? Many thanks