/PointNet_Self_Attention

This repository is an implementation of a class final project that studied the most influential neural architecture for Point Cloud processing called PointNet with the ambition to improve its accuracy using Self-Attention Layers from the famous Transformer architecture.

Primary LanguagePython

PointNet with Self-Attention residual layers

This repository is an implementation of a final course project that studied the most influential neural architecture for Point Cloud processing called PointNet with the ambition to improve its accuracy using Self-Attention Layers from the famous Transformer architecture.

Project Description

 

Some visual results of the Semantic Segmentation algorithm:
Image Description