Shivanand Venkanna Sheshappanavar
Shivanand Venkanna Sheshappanavar
I tried to recreate the robustness test.  MSG+DP model for classification is robust to 1024,768 and 512 input points. However, for 256, and 128 input points the accuracy falls...
## 🐞Describe the bug We have followed the instructions in [Model Scripting](https://coremltools.readme.io/docs/model-scripting) to convert our custom PyTorch model to .mlmodel. Please find our google colab notebook [here](https://colab.research.google.com/drive/1Ha_-zvu0diDdlPnYr8RiEYdD80KBkO9k?usp=sharing) ## Stack Trace...
I just downloaded the code and followed the instructions (installed Python 3.6, Pytorch 1.5 with CUDA 10.2) And when I started training using Multiple GPUs script the training abruptly exited...
Great work BTW!!! I see in the paper you mention about "random picking in the ball" I know the ball query code is in ball_query_gpu.cu file Please point out the...
The accuracy of pointnet++ drops below 60% for input points of 256, 128 and 64. Please let us know how you reproduced the robustness for PointNet++.
Hi, Nice library. Can you please support data loaders to support use of additional features such as normals? Thanks in advance.
Hi, Great work and thanks for releasing the Code. Can you please suggest how one can use additional features such as normals? Thanks in Advance.
Hi If the point cloud has colors. i.e., x,y,z,r,g,b values then how to use those colors while rendering?
I tried to reproduce the results and got following for PointNet++ (Table 4 Row 4 PointNet++)  I am surprised in each category I have higher accuracy. Please let me...
Dear Authors, Can you please verify and confirm if there is a bug? Tensor after grouping: torch.Size([32, 3, 512, 32]) after [Line](https://github.com/guochengqian/openpoints/blob/ee100c81b1d9603c0fc76a3ee4e37d10b2af60ba/models/layers/group.py#L249) After ball querying around an anchor point, the...