Why does the input video need to be square?
lsvagusa opened this issue · 0 comments
lsvagusa commented
I noticed here that it say that the input video MUST be square or 512x512 to be more precise. I've tested the network on multiple videos of various resolutions with varying results. I can't pinpoint a place in the code where this squareness or rather symmetry of the input video is mentioned.
Could somebody help me identify this? Thank you.