Wei Long

@article{pftv22026,
  title={{Progressive Focused Transformer for Image Restoration}},
  author={Long, Wei and Zhang, Jinhua and Zhou, Xingyu and Zhang, Leheng and Gu, Shuhang},
  journal={International Journal of Computer Vision},
  year={2026},
  volume={134},
  number={10},
  pages={458},
  doi={10.1007/s11263-026-03053-1},
  url={https://doi.org/10.1007/s11263-026-03053-1}
}

IDESplat: Iterative Depth Probability Estimation for Generalizable 3D Gaussian Splatting

Wei Long, Haifeng Wu, Shiyin Jiang, Jinhua Zhang, Xinchun Ji, Shuhang Gu

CVPR2026First author

Iterative depth probability refinement improves feed-forward 3D Gaussian reconstruction and novel-view synthesis.

Paper Code

BibTeX

@inproceedings{long2026idesplat,
  author={Long, Wei and Wu, Haifeng and Jiang, Shiyin and Zhang, Jinhua and Ji, Xinchun and Gu, Shuhang},
  title={{IDESplat: Iterative Depth Probability Estimation for Generalizable 3D Gaussian Splatting}},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month={June},
  year={2026},
  pages={33248--33258},
  url={https://openaccess.thecvf.com/content/CVPR2026/html/Long_IDESplat_Iterative_Depth_Probability_Estimation_for_Generalizable_3D_Gaussian_Splatting_CVPR_2026_paper.html}
}

Full-Range Self-Attention framework with bidirectional attention weight selection and attention redistribution connections.

Leveraging negative correlation for Full-Range Self-Attention in Vision Transformers

Wei Long^*, Ziyang Chen^*, Wenting Li, Yongjun Zhang, He Yao, Jiaxin Peng, Zhongwei Cui

Pattern Recognition2026Co-first authorFirst listed 2025

Full-range self-attention uses positive and negative feature correlations to strengthen vision transformers without extra parameters.

Paper Code

BibTeX

@article{fsa2026,
  title={{Leveraging negative correlation for Full-Range Self-Attention in Vision Transformers}},
  author={Long, Wei and Chen, Ziyang and Li, Wenting and Zhang, Yongjun and Yao, He and Peng, Jiaxin and Cui, Zhongwei},
  journal={Pattern Recognition},
  year={2026},
  volume={169},
  pages={111899},
  doi={10.1016/j.patcog.2025.111899},
  url={https://doi.org/10.1016/j.patcog.2025.111899}
}

2025

Progressive Focused Transformer for Single Image Super-Resolution

Wei Long, Xingyu Zhou, Leheng Zhang, Shuhang Gu

CVPR2025First author

Progressively focused attention filters irrelevant tokens to improve both reconstruction quality and computational efficiency.

Paper Code

BibTeX

@inproceedings{long2025progressive,
  author={Long, Wei and Zhou, Xingyu and Zhang, Leheng and Gu, Shuhang},
  title={{Progressive Focused Transformer for Single Image Super-Resolution}},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month={June},
  year={2025},
  pages={2279--2288},
  doi={10.1109/cvpr52734.2025.00218},
  url={https://openaccess.thecvf.com/content/CVPR2025/html/Long_Progressive_Focused_Transformer_for_Single_Image_Super-Resolution_CVPR_2025_paper.html}
}

2024

Overview of MoCha-Stereo, including motif-channel correlation and reconstruction-error motif penalty.

MoCha-Stereo: Motif Channel Attention Network for Stereo Matching

Ziyang Chen^*, Wei Long^*, He Yao^*, Yongjun Zhang, Bingshu Wang, Yongbin Qin, Jia Wu

CVPR2024Co-first author

Motif channel attention preserves geometric structure and refines edge details for stereo matching.

Paper Code

BibTeX

@inproceedings{mocha2024,
  author={Chen, Ziyang and Long, Wei and Yao, He and Zhang, Yongjun and Wang, Bingshu and Qin, Yongbin and Wu, Jia},
  title={{MoCha-Stereo: Motif Channel Attention Network for Stereo Matching}},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month={June},
  year={2024},
  pages={27768--27777},
  doi={10.1109/cvpr52733.2024.02623},
  url={https://openaccess.thecvf.com/content/CVPR2024/html/Chen_MoCha-Stereo_Motif_Channel_Attention_Network_for_Stereo_Matching_CVPR_2024_paper.html}
}

2023

TANet network architecture with AFEM and TAPP modules.

Threshold Attention Network for Semantic Segmentation of Remote Sensing Images

Wei Long, Yongjun Zhang, Zhongwei Cui, Yujie Xu, Xuexue Zhang

IEEE TGRS2023First author

Threshold-based attention models regional dependencies efficiently for semantic segmentation of remote sensing images.

Paper

BibTeX

@article{tanet2023,
  title={{Threshold Attention Network for Semantic Segmentation of Remote Sensing Images}},
  author={Long, Wei and Zhang, Yongjun and Cui, Zhongwei and Xu, Yujie and Zhang, Xuexue},
  journal={IEEE Transactions on Geoscience and Remote Sensing},
  year={2023},
  volume={61},
  pages={1--12},
  doi={10.1109/TGRS.2023.3276081},
  url={https://doi.org/10.1109/TGRS.2023.3276081}
}

Other Co-authored Papers

11 papers

2026

Advanced Pixel Diffusion Model with Guided Sparse Global Refinement

Weiyi You, Jinhua Zhang, Xingyu Zhou, Wei Long, Junyu Lou, Shuhang Gu

arXiv2026

Guided sparse global refinement expands pixel-space diffusion beyond isolated patches while limiting attention cost.

Paper Code

BibTeX

@misc{pixsgr2026,
      title={{Advanced Pixel Diffusion Model with Guided Sparse Global Refinement}},
      author={Weiyi You and Jinhua Zhang and Xingyu Zhou and Wei Long and Junyu Lou and Shuhang Gu},
      year={2026},
      eprint={2609.00798},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2609.00798},
}

Overview of the dual-pass FD-loss post-training framework.

Beyond Token-Level Cross-Entropy: Fréchet Distributional Post-Training for Autoregressive Image Generation

Jinhua Zhang, Yisong Lin, Wei Long, Shuhang Gu

arXiv2026

Fréchet-distance post-training aligns autoregressive image generators with image-level distributional quality.

Paper

BibTeX

@misc{fdloss2026,
      title={{Beyond Token-Level Cross-Entropy: Fr{\'e}chet Distributional Post-Training for Autoregressive Image Generation}},
      author={Jinhua Zhang and Yisong Lin and Wei Long and Shuhang Gu},
      year={2026},
      eprint={2608.00562},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2608.00562},
}

Differentiable Vector Quantization for Rate-Distortion Optimization of Generative Image Compression

Shiyin Jiang, Wei Long, Minghao Han, Zhenghao Chen, Ce Zhu, Shuhang Gu

CVPR · Oral2026

Differentiable vector quantization connects representation learning with rate-distortion optimization for low-bitrate image compression.

Paper Code

BibTeX

@inproceedings{jiang2026differentiable,
  author={Jiang, Shiyin and Long, Wei and Han, Minghao and Chen, Zhenghao and Zhu, Ce and Gu, Shuhang},
  title={{Differentiable Vector Quantization for Rate-Distortion Optimization of Generative Image Compression}},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month={June},
  year={2026},
  pages={14440--14450},
  url={https://openaccess.thecvf.com/content/CVPR2026/html/Jiang_Differentiable_Vector_Quantization_for_Rate-Distortion_Optimization_of_Generative_Image_Compression_CVPR_2026_paper.html}
}

ATD: Improved Transformer With Adaptive Token Dictionary for Image Restoration

Leheng Zhang, Wei Long, Yawei Li, Xingyu Zhou, Xiaorui Zhao, Shuhang Gu

IEEE TPAMI2026

A learned token dictionary brings global image priors and content-adaptive feature grouping to efficient image restoration.

Paper Code

BibTeX

@article{zhang2026atd,
  title={{ATD: Improved Transformer With Adaptive Token Dictionary for Image Restoration}},
  author={Zhang, Leheng and Long, Wei and Li, Yawei and Zhou, Xingyu and Zhao, Xiaorui and Gu, Shuhang},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year={2026},
  volume={48},
  number={7},
  pages={8018--8033},
  doi={10.1109/TPAMI.2026.3669974},
  url={https://doi.org/10.1109/TPAMI.2026.3669974}
}

From Local Windows to Adaptive Candidates via Individualized Exploratory: Rethinking Attention for Image Super-Resolution

Chunyu Meng, Wei Long, Shuhang Gu

ECCV2026

Each token explores its own content-aware attention candidates for efficient image super-resolution.

Paper Code

BibTeX

@inproceedings{iet2026,
  author={Meng, Chunyu and Long, Wei and Gu, Shuhang},
  title={{From Local Windows to Adaptive Candidates via Individualized Exploratory: Rethinking Attention for Image Super-Resolution}},
  editor={Favaro, Paolo and Kukelova, Zuzana and Maki, Atsuto and Rohrbach, Anna and Schindler, Konrad and Tombari, Federico},
  booktitle={Computer Vision -- ECCV 2026},
  series={Lecture Notes in Computer Science},
  volume={17037},
  year={2026},
  publisher={Springer Nature Switzerland},
  address={Cham},
  pages={1--18},
  isbn={978-3-032-37467-7},
  doi={10.1007/978-3-032-37467-7_1},
  url={https://link.springer.com/chapter/10.1007/978-3-032-37467-7_1}
}

Texture Vector-Quantization and Reconstruction Aware Prediction for Generative Super-Resolution

Qifan Li, Jiale Zou, Jinhua Zhang, Wei Long, Xingyu Zhou, Shuhang Gu

ICLR2026First listed 2025

Texture-focused quantization and image-level prediction training support efficient, perceptually rich super-resolution.

Paper Code

BibTeX

@inproceedings{li2026texture,
  author={Li, Qifan and Zou, Jiale and Zhang, Jinhua and Long, Wei and Zhou, Xingyu and Gu, Shuhang},
  title={{Texture Vector-Quantization and Reconstruction Aware Prediction for Generative Super-Resolution}},
  booktitle={International Conference on Learning Representations},
  editor={C. Vondrick and B. Hariharan and C. Raffel and L. Pinto and D. Yang and A. Faust},
  volume={2026},
  year={2026},
  pages={63694--63714},
  url={https://proceedings.iclr.cc/paper_files/paper/2026/hash/67c15da4a9340140c60783d9a175fd3f-Abstract-Conference.html}
}

MVAR: Visual Autoregressive Modeling with Scale and Spatial Markovian Conditioning

Jinhua Zhang, Wei Long, Minghao Han, Weiyi You, Shuhang Gu

ICLR2026First listed 2025

Scale and spatial Markovian conditioning reduces redundant context in visual autoregressive generation.

Paper Code Project

BibTeX

@inproceedings{zhang2026mvar,
  author={Zhang, Jinhua and Long, Wei and Han, Minghao and You, Weiyi and Gu, Shuhang},
  title={{MVAR: Visual Autoregressive Modeling with Scale and Spatial Markovian Conditioning}},
  booktitle={International Conference on Learning Representations},
  editor={C. Vondrick and B. Hariharan and C. Raffel and L. Pinto and D. Yang and A. Faust},
  volume={2026},
  year={2026},
  pages={32889--32917},
  url={https://proceedings.iclr.cc/paper_files/paper/2026/hash/37c6d0bc4d2917dcbea693b18504bd87-Abstract-Conference.html}
}

2025

Rethinking low-light knowledge for pedestrian detection in nighttime conditions

Ziyang Chen, He Yao, Wenting Li, Wei Long, Yongjun Zhang

Engineering Applications of Artificial Intelligence2025

Wavelet-based feature extraction and adaptive knowledge scheduling improve pedestrian detection under low light.

Paper

BibTeX

@article{llknet2025,
  title={{Rethinking low-light knowledge for pedestrian detection in nighttime conditions}},
  author={Chen, Ziyang and Yao, He and Li, Wenting and Long, Wei and Zhang, Yongjun},
  journal={Engineering Applications of Artificial Intelligence},
  year={2025},
  volume={160},
  pages={111979},
  doi={10.1016/j.engappai.2025.111979},
  url={https://doi.org/10.1016/j.engappai.2025.111979}
}

Overall architecture of PGNet, including prior-feature guidance and prior-edge extraction.

Prior Information-Guided Semi-Supervised Semantic Segmentation of Remote Sensing Images

Xiaorong Gan, Wenting Li, Yongjun Zhang, Wei Long, Yujie Lu, Ziyang Chen

IEEE TGRS2025

Feature and edge priors improve pseudo-label reliability and boundaries in semi-supervised remote sensing segmentation.

Paper

BibTeX

@article{pgnet2025,
  title={{Prior Information-Guided Semi-Supervised Semantic Segmentation of Remote Sensing Images}},
  author={Gan, Xiaorong and Li, Wenting and Zhang, Yongjun and Long, Wei and Lu, Yujie and Chen, Ziyang},
  journal={IEEE Transactions on Geoscience and Remote Sensing},
  year={2025},
  volume={63},
  pages={1--16},
  doi={10.1109/TGRS.2025.3571066},
  url={https://doi.org/10.1109/TGRS.2025.3571066}
}

Pipeline of the truncated-short-video training strategy for learning long-range temporal information.

Small Clips, Big Gains: Learning Long-Range Refocused Temporal Information for Video Super-Resolution

Xingyu Zhou, Wei Long, Jingbo Lu, Shiyin Jiang, Weiyi You, Haifeng Wu, Shuhang Gu

arXiv2025

Short-clip training retains long-range temporal information and refocuses attention for efficient video super-resolution.

Paper Code

BibTeX

@misc{lrtivsr2025,
      title={{Small Clips, Big Gains: Learning Long-Range Refocused Temporal Information for Video Super-Resolution}},
      author={Xingyu Zhou and Wei Long and Jingbo Lu and Shiyin Jiang and Weiyi You and Haifeng Wu and Shuhang Gu},
      year={2025},
      eprint={2505.02159},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2505.02159},
}

2024

Overall Multi-Dimensional Manifolds Consistency Regularization framework.

Multi-Dimensional Manifolds Consistency Regularization for semi-supervised remote sensing semantic segmentation

Yujie Lu, Yongjun Zhang, Zhongwei Cui, Wei Long, Ziyang Chen

Knowledge-Based Systems2024

Consistency regularization across image and feature perturbations strengthens remote sensing segmentation with limited labels.

Paper

BibTeX

@article{mdmcr2024,
  title={{Multi-Dimensional Manifolds Consistency Regularization for semi-supervised remote sensing semantic segmentation}},
  author={Lu, Yujie and Zhang, Yongjun and Cui, Zhongwei and Long, Wei and Chen, Ziyang},
  journal={Knowledge-Based Systems},
  year={2024},
  volume={299},
  pages={112032},
  doi={10.1016/j.knosys.2024.112032},
  url={https://doi.org/10.1016/j.knosys.2024.112032}
}