Home

Awesome

[ECCV 2024] Context-Guided Spatial Feature Reconstruction for Efficient Semantic Segmentation

Zhenliang Ni, Xinghao Chen, Yingjie Zhai, Yehui Tang, and Yunhe Wang

🔥 Updates

📸 Overview

<img width="784" alt="cgrseg2" src="https://github.com/nizhenliang/CGRSeg/assets/48742170/eef8c502-599d-48aa-b05b-51a682ac7456">

The overall architecture of CGRSeg. The Rectangular Self-Calibration Module (RCM) is designed for spatial feature reconstruction and pyramid context extraction. The rectangular self-calibration attention (RCA) explicitly models the rectangular region and calibrates the attention shape. The Dynamic Prototype Guided (DPG) head is proposed to improve the classification of the foreground objects via explicit class embedding.

<img width="731" alt="flops" src="https://github.com/nizhenliang/CGRSeg/assets/48742170/2bdf4e0c-d4a7-4b83-b091-394d1ee0afaa">

1️⃣ Results

ADE20K

<img width="539" alt="ade20k" src="https://github.com/nizhenliang/CGRSeg/assets/48742170/98e14385-8f41-417c-84d9-3cc6db0d32c1">

COCO-Stuff-10k

<img width="491" alt="coco" src="https://github.com/nizhenliang/CGRSeg/assets/48742170/9bf2487f-27d6-41d1-8e94-26f3fd994ce0">

Pascal Context

<img width="481" alt="pc" src="https://github.com/nizhenliang/CGRSeg/assets/48742170/d0b3f524-523f-4fc3-a809-691f4617ebb4">

2️⃣ Requirements

3️⃣ Training & Testing

✏️ Reference

If you find CGRSeg useful in your research or applications, please consider giving a star ⭐ and citing using the following BibTeX:

@article{ni2024context,
  title={Context-Guided Spatial Feature Reconstruction for Efficient Semantic Segmentation},
  author={Ni, Zhenliang and Chen, Xinghao and Zhai, Yingjie and Tang, Yehui and Wang, Yunhe},
  journal={arXiv preprint arXiv:2405.06228},
  year={2024}
}