How to implement on a custom video?

[Mzunoven]

Thanks for your great work.

I want to ask how to implement this network on a custom video? Can you give some instructions on how to prepare my video and do prediction?

Thanks!

[abduallahmohamed]

I originally didn't process the data as I used ETH & UCY datasets as it's.
If I were in your place I would:

• Sample the video at fixed periods (0.2 seconds for example).
• Annotate the pedestrians in the video while having a fixed reference.
• Construct the homography matrix that converts from pixel domain to meters(real-world) domain.
• Generate the annotation same way as eth & ucy files are.