pointnet issues

[Part segmentation] wrong concatenation with out5

Dear authors and all, I do want to discuss about the feature concatenation code for part segmentation, specifically, where concatenates local features and global feature in file **pointnet/part_seg/pointnet_part_seg.py**. According to...

hyunjinku

Why is it necessary to add the identity matrix in the transeform.py

2

![image](https://user-images.githubusercontent.com/28957118/68122699-e6b5a100-ff45-11e9-8dd6-51d63597dacc.png)

ShichaoJin

Orthogonality for transform matrix of 'transform_net1' is not counted in the loss calculation.

Dear authors and all, Hi, I just want to double check whether orthogonality for transform matrix of 'transform_net1' (which is 3x3 T-Net matrix for input transform) is counted in your...

hyunjinku

Why the accuracy of train is high and the result of val is poor

1

Hello, I am a newbie in deep learning. I would like to ask, I use the part_seg program to classify a large-scale urban point cloud data set (500m*500m). The training...

yasongguo

C++ implementation

WestbrookZero

I want to use ShapeNet data to classification (make own data for classification)

4

Hello. I am graduate student of Yonsei University, South Korea. I want to convert ShapeNet data for category classification but I think row data format is quite different so I...

hojoonson

Running problems of part segmentation

1

When I run part segmentation, the program reports an error. def get_transform(point_cloud, is_training, bn_decay=None, K = 3): """ Transform Net, input is BxNx3 gray image Return: Transformation matrix of size...

chenMouRen1231