drinkingcoder

Results 5 comments of drinkingcoder

demo and model

Thanks for you attention. We'are now focusing on our CVPR submissions :exploding_head:. The model will be released next week.

Parts are missing in the encoder part

Yes. This option should always be False. We will trim our code that is out of date.

How to fine tune a model to a custom dataset

You could customize the data loader from chairs, or things. But I think the biggest challenge is to obtain the optical flow GT of the your videos.

evaluate_FlowFormer_tile.py has not attribute of MpiSintel_submission

Here is the MpiSintel_submission: https://github.com/drinkingcoder/FlowFormer-Official/blob/21be9d720b24310d1e739d044e812034ab875b33/core/utils/datasets.py#L104

paper is cool, but is there any code available?

Thanks for your attention. We are organizing the code&data for release. Coming soon. :partying_face: