human-activity-dih18

Human Activity Recognition Using Deep Neural Network

Currently use 2d Neural Network to classify the Human Activities

Do the following

1. from UCF 101 on my server select the classes you feel are relevant
2. copy the folders containing these videos into a new folder and edit VIDEO_PATH in config to this folder
3. run extract[3d].py to extract random frames from the videos
4. run fit[3d].py to make a model from the images
5. View online prediction using predict[3d].py . Leave blank to use webcam

To use 3d convnet do the above steps with 3d.py files DTCWT completely uselesss