3D Point Cloud Classification in Python - PointNet Concept and Implementation

Curious about how 3D LiDAR Point Cloud Object classification is done or simply want to know about PointNet? This video shows how a 3D object is classified using PointNet with PyTorch code explanation.

3D point clouds are used extensively by the vast majority of automobile companies (NO, Tesla isn't one of them) and researchers for autonomous vehicles to localize themself and perceive the surrounding environment.

PointNet paper (CVPR 2017) designed a unified architecture that can take unordered point clouds. The architecture is simple but efficient. Both PyTorch and TensorFlow implementations of PointNet are available on GitHub. I tried to explain the architecture and code in the video. I hope it helps you somehow.
--------------------
✅👍📸
Subscribe to the Channel 👉 https://www.youtube.com/channel/UCXeI6BuHPYDd-Po9CZz-bnw?sub_confirmation=1
--------------------
PointNet Paper: https://openaccess.thecvf.com/content_cvpr_2017/papers/Qi_PointNet_Deep_Learning_CVPR_2017_paper.pdf
PyTorch Code: https://github.com/fxia22/pointnet.pytorch
TensorFlow Code: https://github.com/charlesq34/pointnet
--------------------
Chapters
0:00 Intro
0:18 2D Object Classification Brief
1:19 Motivation of 3D Classification and video outline
2:01 PointNet 3D Object Classification Concept
5:38 PointNet Implementation Code
7:42 End
--------------------
Point Cloud Sweater Source: https://sketchfab.com/3d-models/sweatshirt-point-cloud-5e11e9870d91423d928a92be01797cb2
--------------------
#pointnet #pytorch #deeplearning