Looking into the possibility of analyzing simple activities (actors' poses) via single frames
- input_ava.py: Used to generate the input numpy array.
- input_ava_subsample.py: Used to generate the input numpy array with subsampling
- resnet50_AVA.py: The resnet50 run on AVA data set.
- resnet50_mnist.py The resnet50 run on mnist data set.
- resnet101_AVA.py: The resnet101 run on AVA data set.
- resnet101_mnist.py: The resnet101 run on mnist data set.
- label_predict.py: Used to predict labels
python3 input_ava.py --data_path=AVA-II --image_path=/data/ece194n/AVA/keyframes --image_save_path=images
CUDA_VISIBLE_DEVICES=0 python3 resnet50_AVA.py
CUDA_VISIBLE_DEVICES=0 python3 label_predict.py