Wenxuan Wu

But the huber loss in your code doesn't have the cycle-consistency regularization term. Is that correct?

The ShapeNet model will come shortly. Stay tuned.

This repo will no longer be updated. For Semantic Segmentation on ScanNet, Please refer to the code in https://github.com/Xingyi-li-Vince/VI_PointConv, which includes the code to process the dataset and a similar...

Hi, I have tired everything mentioned in the email, including tuning the ratio and the RANSAC threshold. But, none of them works. I still got AUC@20 around 52~53. Cannot reach...

Hi, Thanks for running our code. Unfortunately, I don't have a pretrained model now. But, I can try to get one when I am available. Besides, I want to make...

Hi, If you are only working on self-supervised loss, then you will have to start from step 1. Because the pre-trained weights is from supervised loss, if I remember correctly....

If you use the pretrain weights, then you don't need to fine-tune at all. This model should already give you 0.04~. It depends on if the model converge or not....

Your understanding is correct. But you will have to train more. You can change the https://github.com/DylanWusee/PointPWC/blob/f882e4c396b338d6144bf3d07b335ccaf155bdcf/train_self.py#L124 to init_epoch = 0. So that it can train more. Training on Kitti dataset...