Chen-Suyi/PointRegGPT

[ECCV 2024] PointRegGPT: Boosting 3D Point Cloud Registration using Generative Point-Cloud Pairs for Training, Pytorch implementation.

40
/ 100
Emerging

This project helps improve the accuracy of aligning 3D scans of indoor environments. It takes existing single depth maps of a scene and generates realistic, varied pairs of 3D point clouds that accurately simulate different viewing angles. Engineers, roboticists, or researchers working with 3D scanning and reconstruction will find this useful for training and evaluating their 3D point cloud registration algorithms.

Use this if you need to train or evaluate 3D point cloud registration algorithms for indoor scenes and require more realistic and diverse training data than readily available real-world or simple synthetic datasets.

Not ideal if your primary goal is to perform 3D point cloud registration directly, as this tool is focused on improving the training data for other registration algorithms, not on registration itself.

3D-scanning robotics computer-vision 3D-reconstruction augmented-reality
No Package No Dependents
Maintenance 10 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 6 / 25

How are scores calculated?

Stars

61

Forks

3

Language

Python

License

Apache-2.0

Last pushed

Feb 24, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/Chen-Suyi/PointRegGPT"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.