#Image Restoration

The Cross Aggregation Transformer enhances image restoration by incorporating innovative features such as Rectangle-Window Self-Attention and the Locality Complementary Module, improving on traditional Transformer models. This approach surpasses standard methods by overcoming the limitations of local square windows through horizontal and vertical rectangle attention and cross-window feature aggregation. The Axial-Shift operation further refines window interaction, supporting long-range dependencies. Furthermore, integrating the CNN's inductive bias into the Transformer allows for a hybrid global-local interaction framework, showcasing superior performance in extensive testing compared to recent state-of-the-art methods.

Explore an extensive collection of training and testing codes for leading image restoration models, including USRNet, DnCNN, FFDNet, SRMD, DPSR, MSRResNet, ESRGAN, BSRGAN, SwinIR, VRT, and RVRT. This repository offers capabilities for image enhancement tasks such as super-resolution, deblurring, and denoising leveraging diffusion models and deep learning frameworks. Stay informed with continuous updates and interactive demonstrations, and investigate the application of these methods in practical scenarios via differentiable programming interfaces. It serves as a valuable resource for researchers and developers in the field of visual computing.

The repository provides an extensive overview of methodologies in low-level vision tasks such as image restoration, super-resolution, deraining, and dehazing. It includes noteworthy models like MambaIR, RSDehamba, and MambaCSR, recognized by conferences including ECCV, ACM MM, and NeurIPS 2024. This collection is a valuable asset for those seeking to improve image quality using state-of-the-art state-space models.

Discover how NAFNet transforms image restoration with a simpler and more efficient approach by removing complex nonlinear activations. Achieve exceptional results in deblurring and denoising tasks on established benchmarks like GoPro and SIDD while reducing computational costs.

Explore InstructIR for advanced image restoration guided by human instructions, achieving notable improvements in tasks like denoising, deblurring, deraining, dehazing, and enhancement. Benefit from a dataset designed for text-directed image quality enhancement, suitable for researchers and developers interested in cutting-edge techniques.

This curated collection offers a detailed overview of significant research papers and code from CVPR conferences held in 2024, 2021, and 2020. It focuses on low-level vision tasks, including but not limited to super-resolution, de-raining, dehazing, deblurring, denoising, image restoration and enhancement, as well as inpainting and interpolation. Targeted at researchers and professionals in the field, the repository presents essential tools and knowledge for advancing image processing techniques. By engaging with the content through actions like starring or contributing, users can help drive collaborative development and progress within the community.

Explore this well-structured repository of key papers and codes from ECCV2024 and ECCV2020 that address various low-level vision tasks such as super-resolution, image de-raining, dehazing, and more. The collection supports the research community by encouraging contributions through stars, forks, and PRs, making it a valuable tool for advancing computer vision studies without biased promotion.