This is the github repo of my master's thesis, titled as "Action Recognition and Localization in Videos"
In order to extract 3D features, I use a pretrained 3D ResNet 34 which was available from here
You can find all required packages in requirements.txt file
- Demo file
- Clear code