Squeeze-enhanced axial Transformer

Paper

SeaFormer: Squeeze-enhanced Axial Transformer for Mobile Semantic Segmentation,
Qiang Wan, Zilong Huang, Jiachen Lu, Gang Yu, Li Zhang
ICLR 2023

SeaFormer++: Squeeze-enhanced Axial Transformer for Mobile Visual Recognition,
Qiang Wan, Zilong Huang, Jiachen Lu, Gang Yu, Li Zhang
IJCV 2025

This repository contains the official implementation of SeaFormer.

SeaFormer achieves superior trade-off between performance and latency

The overall architecture of Seaformer

The schematic illustration of the SeaFormer layer

Model Zoo

Image Classification

Classification configs & weights see >>>here<<<.

• SeaFormer on ImageNet-1K

Semantic Segmentation

Segmentation configs & weights see >>>here<<<.

• SeaFormer on ADE20K

• SeaFormer on Cityscapes

BibTeX

@inproceedings{wan2023seaformer,
  title={Seaformer: Squeeze-enhanced axial transformer for mobile semantic segmentation},
  author={Wan, Qiang and Huang, Zilong and Lu, Jiachen and Gang, YU and Zhang, Li},
  booktitle={International Conference on Learning Representations (ICLR)},
  year={2023}
}

@article{wan2025seaformer++,
  title={SeaFormer++: Squeeze-enhanced axial transformer for mobile visual recognition},
  author={Wan, Qiang and Huang, Zilong and Lu, Jiachen and Yu, Gang and Zhang, Li},
  journal={International Journal of Computer Vision (IJCV)},
  year={2025}
}

Acknowledgment

Thanks to previous open-sourced repo:
TopFormer
mmsegmentation
pytorch-image-models