video SR

*1.TDAN: Temporally-Deformable Alignment Network for Video Super-Resolution(做不了)
pytorch-0.3.1
an NVIDIA 1080TI GPU
github:https://github.com/YapengTian/TDAN-VSR-CVPR-2020
ps:
由于pytorch-0.3.1 做不了
*2.Zooming Slow-Mo: Fast and Accurate One-Stage Space-Time Video Super-Resolution.(尝试失败, 3090不支持pytorch1.4,导致里面的DCNv2版本没法编译)
PyTorch >= 1.1
(一张Nvidia Titan XP GPU.)
github:https://github.com/Mukosame/Zooming-Slow-Mo-CVPR-2020
ps:
由于之前接触过一点点这个项目, 里面有一个可变形模块DCNv2。当时这个模块需要pytorch-1.4
借用的是https://github.com/CharlesShang/DCNv2
后来有人改进了这个模块可以在3090上编译https://github.com/MatthewHowe/DCNv2(失败)
*3.Video Super-resolution with Temporal Group Attention (TGA)(不能)(先看论文)
PyTorch 1.1
(8 Nvidia Tesla V100 GPUs)
github:https://github.com/junpan19/VSR_TGA
ps:
8块v100,不知道一块3090会跑几天
RuntimeError: CUDA out of memory. Tried to allocate 648.00 MiB (GPU 0; 23.70 GiB total capacity; 21.72 GiB already allocated; 169.25 MiB free; 22.17 GiB reserved in total by PyTorch)
*4.Space-Time-Aware Multi-Resolution Video Enhancement(显存不够)(数据集特别大)
PyTorch >= 1.0.0
NVIDIA Tesla V100 GPUs(具体不知道几块v100)
github:https://github.com/alterzero/STARnet
RuntimeError: CUDA out of memory. Tried to allocate 280.00 MiB (GPU 0; 23.70 GiB total capacity; 21.96 GiB already allocated; 113.25 MiB free; 22.23 GiB reserved in total by PyTorch)
*5.FISR: Deep Joint Frame Interpolation and Super-Resolution with A Multi-scale Temporal Loss(做不了)
Tensorflow 1.13
NVIDIA TITAN Xp GPU
github:https://github.com/JihyongOh/FISR
*6.Video Face Super-Resolution with Motion-Adaptive Feedback Cell(不能做, 没代码)
one Titan X Pascal GPU.
*7.JSI-GAN: GAN-Based Joint Super-Resolution and Inverse Tone-Mapping with Pixel-Wise Task-Specific Filters for UHD HDR Video(没代码)
Tensorflow 1.13
*8.VESR-Net: The Winning Solution to Youku Video Enhancement and Super-Resolution Challenge(无代码)
9.STVUN: Deep Space-Time Video Upsampling Networks(貌似比FISR好)(可以试试)(没有train代码,只有test)
Pytorch=1.2
Nvidia Geforce Titan X
github:https://github.com/JaeYeonKang/STVUN-Pytorch
10.MuCAN: Multi-Correspondence Aggregation Network for Video Super-Resolution(没train代码,只有test)
eight NVIDIA GeForce GTX 1080TiGPUs
github:https://github.com/Jia-Research-Lab/Simple-SR
11.Video Super-Resolution with Recurrent Structure-Detail Network (RSDN)(暂无train带代码)
pytorch 1.1
8 Nvidia Tesla V100 GPUs for training.
github:https://github.com/junpan19/RSDN
12.Neural Supersampling for Real-time Rendering(还未完工)
github:https://github.com/IMAC-projects/NSRR-PyTorch
13.Video Restoration with Enhanced Deformable Convolutional Networks(正在尝试)
8 NVIDIA Titan Xp GPUs
github:https://github.com/xinntao/EDVR