POSTECH-CVLab/SCNeRF
A PyTorch implementation of a self-calibrating neural radiance field system that jointly learns scene geometry and camera parameters without calibration objects.
Not currently ranked — collecting fresh signals.
star history
This research project implements a self-calibration algorithm for generic cameras with non-linear distortions by combining Neural Radiance Fields with camera model estimation. The system uses a pinhole model with radial distortion and a learnable noise model to handle arbitrary non-linear camera distortions. It incorporates both photometric consistency (via NeRF) and geometric constraints through a projected ray distance loss function, enabling camera calibration without calibration objects.
Frequently asked
- What is POSTECH-CVLab/SCNeRF?
- A PyTorch implementation of a self-calibrating neural radiance field system that jointly learns scene geometry and camera parameters without calibration objects.
- Is SCNeRF open source?
- Yes — POSTECH-CVLab/SCNeRF is open source, released under the MIT license.
- What language is SCNeRF written in?
- POSTECH-CVLab/SCNeRF is primarily written in Python.
- How popular is SCNeRF?
- POSTECH-CVLab/SCNeRF has 471 stars on GitHub.
- Where can I find SCNeRF?
- POSTECH-CVLab/SCNeRF is on GitHub at https://github.com/POSTECH-CVLab/SCNeRF.