PointCloudLib icon indicating copy to clipboard operation
PointCloudLib copied to clipboard

Jittor Library for Point Cloud Processing

Results 9 PointCloudLib issues
Sort by recently updated
recently updated
newest added

我训练完jittor版的pointnet,速度很快,精度很高,作者很强。 但是怎么可视化分类或者分割的结果呢?

```bash Traceback (most recent call last): File "train_cls.py", line 18, in from data_utils.kpconv_loader import KPConvLoader File "/home/jiahui/Desktop/ComputerGraph/pointnet/PointCloudLib/data_utils/kpconv_loader.py", line 10, in import cpp_wrappers.cpp_neighbors.radius_neighbors as cpp_neighbors ModuleNotFoundError: No module named 'cpp_wrappers.cpp_neighbors.radius_neighbors' ```...

I found that your pointnet model is conv1d, but the origin is MLP. And the origin has the T-net, but your model didn't have the T-net? Why did you design...

Hi, @MenghaoGuo , when I ran this code, `CUDA_VISIBLE_DEVICES=0 python train_cls.py --model pointconv` I got the following mistakes, `Traceback (most recent call last): File "train_cls.py", line 155, in train(net, optimizer,...

Hi, all, I can train the pointnet and pointnet++ model for the part segmentation task successfully. Here is my command: `python train_partseg.py --model pointnet`. However, when I train the pointconv...

when I run the code you provide in PCT, the acc is only 87%, why ?

Hi, all, In [train_partseg.py](https://github.com/Jittor/PointCloudLib/blob/master/train_partseg.py#L86), lr_scheduler is applied. However, this argument is assigned with value around 14,000 (for ShapeNet Part dataset), which is passed to this [line](https://github.com/Jittor/PointCloudLib/blob/master/misc/utils.py#L17): `lr_decay = self.lr_decay **...

Hi, all, Could you provide the trained models mentioned [here](https://github.com/Jittor/PointCloudLib#%E5%B7%B2%E7%BB%8F%E5%AE%9E%E7%8E%B0%E7%9A%84%E6%A8%A1%E5%9E%8B)? Thanks~