pointnet_pytorch

PointNet.pytorch

Input

(Image from http://web.stanford.edu/~ericyi/project_page/part_annotation/index.html)

• Segmentation model Input Shape : (1, 3, N)
• Classification model Input Shape : (batch, 3, N)

Output

• Segmentation model output shape
  • pred shape : (1, N, Seg_class)
  • trans shape : (1, 3, 3)
• Classification model output shape
  • pred shape : (batch, 16)
  • trans shape : (batch, 3, 3)

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample image,

$ python3 pointnet_pytorch.py

You can specify the "class" by specifying after the --choice_class option.
The class is selected from airplane, bag, cap, car, chair, earphone, guitar, knife, lamp, laptop, motorbike, mug, pistol, rocket, skateboard, table.

$ python3 pointnet_pytorch.py --choice-class CLASS

If you want to specify the input point, put the .pts file path after the --input option.
You can use --savepath option to change the name of the output file to save.

$ python3 pointnet_pytorch.py --input POINT_FILE_PATH --savepath SAVE_IMAGE_PATH

Reference

• PointNet.pytorch

Framework

Pytorch

Model Format

ONNX opset=11

Netron

airplane_100.onnx.prototxt bag_100.onnx.prototxt cap_100.onnx.prototxt car_100.onnx.prototxt chair_100.onnx.prototxt earphone_100.onnx.prototxt guitar_100.onnx.prototxt knife_100.onnx.prototxt lamp_100.onnx.prototxt laptop_100.onnx.prototxt motorbike_100.onnx.prototxt mug_100.onnx.prototxt pistol_100.onnx.prototxt rocket_100.onnx.prototxt skateboard_100.onnx.prototxt table_100.onnx.prototxt cls_model_100.onnx.prototxt