Task-driven Image Fusion with Learnable Fusion Loss

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025. (Highlight)

Haowen Bai, Jiangshe Zhang*, Zixiang Zhao*, Yichen Wu, Lilun Deng, Yukun Cui, Tao Feng, Shuang Xu

Paper | ArXiv | Code

• A task-driven image fusion framework that employs a learnable fusion loss guided by downstream task objectives through meta-learning, enabling adaptive fusion optimization for improved performance in downstream tasks.

Deep Unfolding Multi-modal Image Fusion Network via Attribution Analysis

IEEE Transactions on Circuits and Systems for Video Technology (TCSVT), 35(4), pp. 3498-3511, 2025.

Haowen Bai, Zixiang Zhao, Jiangshe Zhang*, Baisong Jiang, Lilun Deng, Yukun Cui, Shuang Xu, Chunxia Zhang

Paper | ArXiv | Code

• An attribution-guided fusion framework that optimizes multi-modal image synthesis for semantic segmentation via unfolding networks and adaptive loss design, prioritizing task-critical features through attribution maps.

ReFusion: Learning Image Fusion from Reconstruction with Learnable Loss Via Meta-Learning

International Journal of Computer Vision (IJCV), pp.1-23, 2024.

Haowen Bai, Zixiang Zhao*, Jiangshe Zhang*, Yichen Wu, Lilun Deng, Yukun Cui, Baisong Jiang, Shuang Xu

Paper | ArXiv | Code

• Propose a meta-learning-based image fusion framework that dynamically optimizes task-specific loss functions through source image reconstruction to preserve optimal source information.

Simultaneous Automatic Picking and Manual Picking Refinement for First-Break

IEEE Transactions on Geoscience and Remote Sensing (TGRS), 62, pp. 1-12, 2024.

Haowen Bai, Zixiang Zhao, Jiangshe Zhang*, Yukun Cui, Chunxia Zhang*, Zhenbo Guo, Yongjun Wang

Paper | ArXiv

• Propose a novel approach that jointly optimizes label refinement and first-break picking performance in microseismic data by integrating a latent variable representation of true first-break times into a probabilistic model.

Image fusion via vision-language model

Forty-first International Conference on Machine Learning (ICML), 2024.

Zixiang Zhao, Lilun Deng, Haowen Bai, Yukun Cui, Zhipeng Zhang, Yulun Zhang, Haotong Qin, Dongdong Chen, Jiangshe Zhang, Peng Wang, Luc Van Gool

Project Page | Paper | ArXiv | Code

• Introduce a novel fusion paradigm, for the first time, utilizing explicit textual information to guide image fusion.

Equivariant multi-modality image fusion

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024.

Zixiang Zhao, Haowen Bai, Jiangshe Zhang, Yulun Zhang, Kai Zhang, Shuang Xu, Dongdong Chen, Radu Timofte, Luc Van Gool

Paper | ArXiv | Code

• Propose a novel end-to-end self-supervised fusion algorithm based on the equivariant sensing and imaging prior.

DDFM: denoising diffusion model for multi-modality image fusion

IEEE/CVF International Conference on Computer Vision (ICCV), 2023. (Oral)

Zixiang Zhao, Haowen Bai, Yuanzhi Zhu, Jiangshe Zhang, Shuang Xu, Yulun Zhang, Kai Zhang, Deyu Meng, Radu Timofte, Luc Van Gool

Paper | ArXiv | Code

• Propose a novel fusion algorithm based on the denoising diffusion sampling model.

CDDfuse: Correlation-Driven Dual-branch feature decomposition for multi-modality image fusion

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023.

Zixiang Zhao, Haowen Bai, Jiangshe Zhang, Yulun Zhang, Shuang Xu, Zudi Lin, Radu Timofte, Luc Van Gool

Paper | ArXiv | Code

• Propose a Correlation-Driven feature Decomposition Fusion (CDDFuse) network for multi-modality image fusion.

Deep convolutional sparse coding networks for interpretable image fusion

IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2023.

Zixiang Zhao, Jiangshe Zhang, Haowen Bai, Yicheng Wang, Yukun Cui, Lilun Deng, Kai Sun, Chunxia Zhang, Junmin Liu, Shuang Xu

Paper | Code

• Gave three deep convolutional sparse coding networks for interpretable image fusion via unfolding the iterative shrinkage and thresholding algorithm.