ActivityDetection

I did this project in 2018. I used a preTrain resnet network to learn this network with Kinect Activity Dataset. The dataset has some frames videos as each activity every frame has one person with 25 joints with X, Y and Z. The network was learned as an image, and coordinated as RGB, and the number of frames is the image’s width. I used Python, PyTorch, scikit-learn, NumPy, Matplotlib and TensorBoard in PyTorch. For more detail of TensorBoard in PyTorch https://github.com/yunjey/pytorch-tutorial/tree/master/tutorials/04-utils/tensorboard