DualPM/DualPM_Paper

Training, inference and some benchmark code for DualPM presented at CVPR 2025 - Highlight

32
/ 100
Emerging

This project helps computer vision researchers and 3D graphics professionals reconstruct the true 3D shape and orientation of deformable objects from a single image. It takes an image of an object and outputs a pair of point maps representing the object's geometry in both its observed pose and a standardized, canonical pose. This enables detailed 3D modeling and analysis of objects like human bodies or animals from 2D photos.

Use this if you need to precisely recover the 3D geometry and pose of deformable objects from individual images for applications like animation, medical imaging, or virtual reality.

Not ideal if you are working with rigid objects, require real-time processing without offline feature extraction, or do not have access to CUDA-enabled hardware.

3D-reconstruction computer-vision deformable-objects graphics-research pose-estimation
No Package No Dependents
Maintenance 6 / 25
Adoption 6 / 25
Maturity 15 / 25
Community 5 / 25

How are scores calculated?

Stars

17

Forks

1

Language

Python

License

BSD-3-Clause

Last pushed

Nov 04, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/DualPM/DualPM_Paper"

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