hanna-xu/U2Fusion

Code of U2Fusion: a unified unsupervised image fusion network for multiple image fusion tasks, including multi-modal, multi-exposure and multi-focus image fusion.

What it solves

U2Fusion is designed to handle multiple image fusion tasks in a single, unified network. It addresses the challenge of combining information from different images of the same scene—such as visible and infrared (VIS-IR) images, medical images, images with different exposures, and images with focus levels—into one high-quality image.

How it works

It uses an unsupervised image fusion network. This means it can learn to merge images without needing paired ground-truth labels for training, which is a critical advantage for image fusion tasks where perfect reference images are often unavailable.

Who it’s for

Researchers and developers working in computer vision and image processing, specifically those needing to merge multi-modal or multi-exposure images for better scene understanding.

Highlights

  • Unified architecture: A single network can handle VIS-IR, medical, multi-exposure, and multi-focus fusion.
  • Unsupervised learning: Eliminates the need for labeled training data.
  • PyTorch implementation available: The project provides a separate repository for PyTorch users.

Related

  • Project
    Linfeng-Tang/SwinFusion

    SwinFusion is a Swin Transformer-based framework for general image fusion that combines information from multiple source images across various domains, such as infrared, visible, and medical imaging.

  • Project
    Linfeng-Tang/Image-Fusion

    A comprehensive collection of research papers and code implementations for image fusion, covering multi-modal, digital photography, and remote sensing image combination techniques.

  • Project
    nagadomi/nunif

    A collection of PyTorch-based tools for image super-resolution, 2D-to-3D video conversion for VR, and image quality assessment for dataset filtering.

  • Project
    SkyworkAI/UniPic

    A unified multimodal model series for image editing, generation, and understanding, featuring high-speed multi-image editing and autoregressive modeling.

  • Project
    Linfeng-Tang/SeAFusion

    A semantic-aware real-time network that fuses infrared and visible images to optimize them for high-level vision tasks like object detection and semantic segmentation.