Deep learning pipeline for 6D pose estimation (translation and rotation) of rigid objects from RGB-D images, based on a modified PointNet++ architecture. It leverages both geometric and image features for high-accuracy results tailored for robotic manipulation tasks.
PyTorch benchmark of PointNet, PointNet++ (SSG/MSG), and PointCNN for outdoor LiDAR semantic segmentation on Paris-Lille-3D and Toronto-3D.
Interactive PointNet++ workshop for point cloud deep learning. Train a deep learning model to classify 3D shapes, visualize features. GPU-optimized Jupyter notebook.
Semantic Segmentation of an Industrial 3D Point Cloud
PointNet++ primitive-shape classification with APS-to-YCB synthetic-to-real generalisation and ModelNet40 evaluation.
First annotated 3D LiDAR data analysis for broccoli and mushroom moisture prediction. Combines geometric point clouds with spectral intensity. Implements PointNet++ models integrating structure and spectra for improved classification accuracy.
Hindi air-writing gesture recognition using PointNet++ and Leap Motion for 3D hand gesture classification
To associate your repository with the pointnet-plus-plus topic, visit your repo's landing page and select "manage topics."