rcan 开源项目推荐（17 个）

XPixelGroup /BasicSR

开源图像与视频修复工具箱，用于超分辨率、去噪、去模糊等。目前包含 EDSR、RCAN、SRResNet、SRGAN、ESRGAN、EDVR、BasicVSR、SwinIR、ECBSR 等。还支持 StyleGAN2、DFDNet。

LoSealL /VideoSuperResolution

A collection of state-of-the-art video or single-image super-resolution architectures, reimplemented in tensorflow.

Coloquinte /torchSR

Super Resolution datasets and models in Pytorch

kozistr /rcan-tensorflow

Image Super-Resolution Using Very Deep Residual Channel Attention Networks Implementation in Tensorflow

george-gca /sr-pytorch-lightning

在 PyTorch Lightning 中实现的超分辨率模型

orange-fritters /fast-mri

[FastMRI Challenge] E2E-VarNet + RCAN Combination for MRI Reconstruction

niliusha123 /SR_LIIF

Quickly process image super-resolution reconstruction at any scale ratio

fangvv /CDNSR

Code for paper "Classification-based Dynamic Network for Efficient Super-Resolution"

RobotRegistryFoundation /rcan-spec

RCAN (Robot Communication and Addressing Network): open protocol spec and conformance suite. Live versions at rcan.dev/compatibility.

Assem-ElQersh /General-SISR-Framework

Production-grade Single Image Super-Resolution framework in pure PyTorch

RobotRegistryFoundation /rcan-py

Python SDK for RCAN (Robot Communication and Addressing Network). Live versions at rcan.dev/compatibility.

TheOnlySG /Super-Resolution-Model

SIH 2026 · Problem Statement 26142 End-to-end 4× Multispectral Satellite Imagery Super-Resolution, Land Cover Segmentation, and Scientific Quality Validation.

RobotRegistryFoundation /rcan-ts

TypeScript SDK for RCAN (Robot Communication and Addressing Network). Live versions at rcan.dev/compatibility.

RobotRegistryFoundation /robot-md

ROBOT.md: a single self-describing file so any agent can safely operate your robot.

Diouo /MSRN-RCAN-SRGAN

SRGAN named MSCAR-GAN combined with EDSR, RCAN, MSRN

ash1ra /RCAN

RCAN (Residual Channel Attention Network) model for SISR (Single Image Super-Resolution) task using PyTorch

mhruiz /tf-super-resolution-RCAN

Tensorflow 2.4 implementation of the RCAN Super-Resolution architecture, trained and validated on DIV2K datasets, and tested on recognized image sets (Set5, Set14, Urban100, Manga109)