whigg/pc_autoencoder

Auto Encoder for 3D Point Clouds using ModelNet40 Dataset

Auto Encoder for 3D Point Clouds

Network Structure:

Code:

Steps to train the auto-encoder:

1. Download ModelNet40 Dataset [Link]
2. Clone repository.
3. Extract the zip file and copy modelnet40_ply_hdf5_2048 folder to pc_autoencoder/data.
4. python pointnet_autoencoder_train.py --mode train

Steps to test the auto-encoder:

1. Download dataset as given in training steps.
2. Download weights for the trained network. [Link]
3. python pointnet_autoencoder_train.py --mode test

Visualise the Dataset: python show_pc.py idx idx: Index of Point Cloud in ModelNet40 Dataset.

Results:

Red colored point clouds are input to the network and blue point clouds are the output.

[Note: A translation has been applied to blue point clouds during testing for a better visualisation purpose.]

Additional Results:

References:

1. PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation [Link]
2. PCN: Point Completion Network [Link]