ECCV2022-RIFE
ECCV2022-RIFE copied to clipboard
hzwer
ECCV2022 - Real-Time Intermediate Flow Estimation for Video Frame Interpolation
有关模型训练问题
作者你好，我利用光流预测模型生成的光流图去约束rife生成的光流图，从结果上来看rife生成的光流图整体不错但是在细节方面不是很准确，我目前采用l1对光流计算损失，请问我可以采用其他什么损失能够提高生成光流图的准确性呢？
Hi, could you please describe what exactly is done and what's the output of encode in IFNet.forward() ``` f0 = self.encode(img0[:, :3]) f1 = self.encode(img1[:, :3]) ``` It outputs tensor...
Hi, I saw the issue #278 , however In my opinion rife_m model provided in older version lacks the quality which is provided in latest versions. My question is how...
Hi, I just tried to use RIFE for the first time and the output video is simply broken. I followed the instructions from the README and ran the command `python3...
After cloning the repository and running `pip3 install -r requirements.txt`, I get the following errors: ``` ERROR: Ignored the following versions that require a different python version: 1.21.2 Requires-Python >=3.7,=3.7,=3.7,=3.7,=3.7,...
I run the command: python inference_video.py --exp=2 --video=test.mp4 But it only uses my cpu, how to direct the load to the gpu? I have a 1660 super
I'm using RIFE via Flowframes, how do I resolve this error? "Error: Failed to retrieve frame 0 with error: Resize error: Resize error 3074: unrecognized color primaries (1/1/0 => 0/1/0)....
flow, mask, merged = self.flownet(torch.cat((imgs, gt), 1), scale=scale, training=training) **loss_l1** = (merged[2] - gt).abs().mean() **loss_smooth** = self.sobel(flow[2], flow[2]*0).mean() # loss_vgg = self.vgg(merged[2], gt) if training: self.optimG.zero_grad() **loss_G = loss_cons +...
I have a question about arbitrary-timestep frame interpolation. As explained in the paper, this was achieved by the model RIFEm by directly inputting _t_ to the IFNet (which creates a...
