yassouali/CCT
Implementation of a CVPR 2020 paper on semi-supervised semantic segmentation using cross-consistency training with deep neural networks.
Not currently ranked — collecting fresh signals.
star history
This repository implements the Cross-Consistency Training (CCT) method for semi-supervised semantic segmentation. It adapts consistency training frameworks to the dense prediction setting of segmentation, enforcing consistency over encoder outputs rather than inputs. The approach uses perturbations and can incorporate weak labels and multi-domain pixel-level labels alongside unlabeled data.
Frequently asked
- What is yassouali/CCT?
- Implementation of a CVPR 2020 paper on semi-supervised semantic segmentation using cross-consistency training with deep neural networks.
- Is CCT open source?
- Yes — yassouali/CCT is open source, released under the MIT license.
- What language is CCT written in?
- yassouali/CCT is primarily written in Python.
- How popular is CCT?
- yassouali/CCT has 411 stars on GitHub.
- Where can I find CCT?
- yassouali/CCT is on GitHub at https://github.com/yassouali/CCT.