wentaoyuan/pcn
A TensorFlow implementation of a deep learning model that completes partial 3D point cloud scans into dense, full shapes.
Not currently ranked — collecting fresh signals.
star history
PCN is a learning-based shape completion method that maps partial point clouds to complete ones without voxelization. The network is trained on ShapeNet data and directly outputs dense point clouds. It uses a decoder architecture with fully connected layers followed by a folding module to generate the completed shape from a latent feature vector extracted from the input.
Frequently asked
- What is wentaoyuan/pcn?
- A TensorFlow implementation of a deep learning model that completes partial 3D point cloud scans into dense, full shapes.
- Is pcn open source?
- Yes — wentaoyuan/pcn is open source, released under the MIT license.
- What language is pcn written in?
- wentaoyuan/pcn is primarily written in Python.
- How popular is pcn?
- wentaoyuan/pcn has 482 stars on GitHub.
- Where can I find pcn?
- wentaoyuan/pcn is on GitHub at https://github.com/wentaoyuan/pcn.