SkyeLu comments

Results 29 comments of


                                            SkyeLu

你好

Hi，训练是在一张2080ti上完成的，跑完完整的实验（包括masa-rec和masa这两个模型）大约需一天半时间。

你好

可能跟你的环境有关？没记错的话，我当时的环境是pytorch1.4.0，torchvision0.5.0。

/python3.10/site-packages/psbody/mesh/serialization/serialization.py", line 435, in load_from_ply raise SerializationError(e) psbody.mesh.errors.SerializationError: Failed to open PLY file.

Guys, I met the same problem, then it turned out that I forgot to put the necessary files in `./vocaset/templates`. Maybe some of you made the same mistake.

Web Interface

Wow, thanks for your amazing work! I will add your link to our README file.

issue of compute_flow_vimeo.py

Hi, you may refer to this issue: https://github.com/sniklaus/pytorch-liteflownet/issues/21

Question on ablation study

Hi, thanks for your interest in our work. As mentioned in the appendix of our paper, the main difference of Model 1 and the best baseline model is the flow...

issue of missing checkpoints/finetuned-liteflownet-epoch1.pkl

1. Hi, the finetuned liteflownet checkpoint can be downloaded from repo of RIFE: https://drive.google.com/file/d/1XS0yR0ZGSFxf7gSVuQ9ahj658acEi2fF/view 2. Yes. Change these two lines: https://github.com/dvlab-research/VFIformer/blob/b81f2ae49ba7774dfae0607ff7b05bbb9aaf1b06/compute_flow_vimeo.py#L378-L379

issue of code

Hi, please install timm package via "pip install timm".

Missing flow .npy file

Hi, thanks for raising this issue, we have updated the `Prepare Dataset` part in the README file, please follow the instruction to compute the GT flows by using lite-flownet.