SHI-Labs/Neighborhood-Attention-Transformer
A transformer architecture for computer vision that uses localized attention mechanisms, published at CVPR 2023.
Not currently ranked — collecting fresh signals.
star history
Neighborhood Attention Transformer (NAT) and its dilated variant (DNAT) are vision transformer architectures that replace global self-attention with localized neighborhood attention for improved efficiency. The models achieve state-of-the-art performance on multiple computer vision benchmarks including instance segmentation, semantic segmentation, and panoptic segmentation on ADE20K, Cityscapes, and COCO datasets. The implementation includes PyTorch models and a custom CUDA extension (NATTEN) for accelerated neighborhood attention computation.
Frequently asked
- What is SHI-Labs/Neighborhood-Attention-Transformer?
- A transformer architecture for computer vision that uses localized attention mechanisms, published at CVPR 2023.
- Is Neighborhood-Attention-Transformer open source?
- Yes — SHI-Labs/Neighborhood-Attention-Transformer is open source, released under the MIT license.
- What language is Neighborhood-Attention-Transformer written in?
- SHI-Labs/Neighborhood-Attention-Transformer is primarily written in Python.
- How popular is Neighborhood-Attention-Transformer?
- SHI-Labs/Neighborhood-Attention-Transformer has 1.2k stars on GitHub.
- Where can I find Neighborhood-Attention-Transformer?
- SHI-Labs/Neighborhood-Attention-Transformer is on GitHub at https://github.com/SHI-Labs/Neighborhood-Attention-Transformer.