Two Stream CNN for Human Activity Recognition in Video (Temporal + Spatial)

Here we use Temporal Stream for Optical Flow and Spatial Stream for RGB Frames

1. Download UCF-101 Dataset and place it inside Dataset before start training.
2. To start Training run Video Classification With Two Stream jupyter notebook file
3. You can directly download the pre-trained model from https://drive.google.com/file/d/16Ck9tsbxCzxj9N6FGYW1n3_EKyNZiArf/view?usp=sharing
4. Put your test video inside TestVideos folder and run Prediction jupyter notebook for prediction