NTIRE 的全称为New Trends in Image Restoration and Enhancement Challenges,即“图像复原和增强挑战中的新趋势”,是CVPR(IEEE Conference on Computer Vision and Pattern Recognition)举办的极具影响力的计算机视觉底层任务比赛,主要涉及的研究方向有:图像超分辨率、图像去噪、去模糊、去摩尔纹、重建和去雾等。
其中在2022年,CVPR开展的NTIRE相关挑战有:
光谱恢复(spectral recovery);
光谱去马赛克(spectral demosaicing);
感知图像质量评估(perceptual image quality assessment);
修复(inpainting);
夜间摄影渲染(night photography rendering);
高效超分辨率(efficient super-resolution);
学习超分辨率空间(learning the super-resolution space);
压缩视频的超分辨率和视频增强(super-resolution and quality enhancement of compression video);
高范围动态(high dynamic range);
双目超分辨率(stereo super-resolution);
真实世界超分辨率(burst super resolution)。
- RFDN:GitHub - njulj/RFDN: Residual Feature Distillation Network for Lightweight Image Super-Resolution
- FightingCV 代码库:GitHub - xmu-xiaoma666/External-Attention-pytorch: 🍀 Pytorch implementation of various Attention Mechanisms, MLP, Re-parameter, Convolution, which is helpful to further understand papers.⭐⭐⭐
- ECBSR:GitHub - xindongzhang/ECBSR: Edge-oriented Convolution Block for Real-time Super Resolution on Mobile Devices, ACM Multimedia 2021
- RLFN:GitHub - bytedance/RLFN: Winner of runtime track in NTIRE 2022 challenge on Efficient Super-Resolution
- EDSR:GitHub - sanghyun-son/EDSR-PyTorch: PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)NTIRE 2022 Challenge on Stereo Image Super-Resolution: Methods and ResultsGitHub - sanghyun-son/EDSR-PyTorch: PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)