How can we extract optical flow features using Flowformer on own video dataset?

[prasaddev9]

Hi @drinkingcoder .

Thank you for your great work!

What are the necessary steps that we can extract optical flow features using Flowformer on own video dataset

[drinkingcoder]

What do you mean about optical flow features? Maybe you can extract features from the cost memory encoded by our AGT cost encoder?