search

MichiganCOG / ViP

Video Platform for Action Recognition and Object Detection in Pytorch
MIT License
220 stars 37 forks source link

Add YC2-BB dataset #31

Closed natlouis closed 5 years ago

natlouis commented 5 years ago

For ticket #4.

This performs the evaluation on the validation set. With the pretrained weights, the same accuracy score is achieved.

Requires a big change in train.py and eval.py because the model has several input arguments. So I'll leave this as WIP until we figure that out.

natlouis commented 5 years ago

Needs #36 to be merged before completing. Because all frames features need to be input at once.

natlouis commented 5 years ago

Focus on changes to train.py and eval.py and check that the model runs fine.

Just training a model now to retrieve our "trained" model's accuracy.