SysCV/transfiner
A transformer-based instance segmentation model achieving state-of-the-art high-quality mask predictions using a quadtree structure.
Not currently ranked — collecting fresh signals.
star history
Mask Transfiner is a PyTorch implementation of a high-quality instance segmentation system published at CVPR 2022. The method uses an efficient transformer architecture with quadtree-based predictions to generate precise instance masks with fine boundaries. Built on top of the detectron2 framework, it demonstrates improved mask and boundary AP on benchmarks including COCO, Cityscapes, and BDD100k compared to standard approaches.
Frequently asked
- What is SysCV/transfiner?
- A transformer-based instance segmentation model achieving state-of-the-art high-quality mask predictions using a quadtree structure.
- Is transfiner open source?
- Yes — SysCV/transfiner is open source, released under the Apache-2.0 license.
- What language is transfiner written in?
- SysCV/transfiner is primarily written in Python.
- How popular is transfiner?
- SysCV/transfiner has 546 stars on GitHub.
- Where can I find transfiner?
- SysCV/transfiner is on GitHub at https://github.com/SysCV/transfiner.