facebookresearch/Mask2Former
A transformer neural network architecture for universal image segmentation across panoptic, instance, and semantic tasks.
Not currently ranked — collecting fresh signals.
star history
Mask2Former is a computer vision model using masked-attention transformers for universal image segmentation. It achieves state-of-the-art performance on panoptic, instance, and semantic segmentation tasks using a single unified architecture. The project supports major benchmarks including ADE20K, Cityscapes, COCO, and Mapillary Vistas, with trained models and a web demo available through Hugging Face Spaces.
Frequently asked
- What is facebookresearch/Mask2Former?
- A transformer neural network architecture for universal image segmentation across panoptic, instance, and semantic tasks.
- Is Mask2Former open source?
- Yes — facebookresearch/Mask2Former is open source, released under the MIT license.
- What language is Mask2Former written in?
- facebookresearch/Mask2Former is primarily written in Python.
- How popular is Mask2Former?
- facebookresearch/Mask2Former has 3.4k stars on GitHub.
- Where can I find Mask2Former?
- facebookresearch/Mask2Former is on GitHub at https://github.com/facebookresearch/Mask2Former.