xieyulai / UAVHuman_For_TE-GCN

Generating splited UAV-Human Pose datasets (XSUB1 and XSUB2) for [TE-GCN](https://github.com/xieyulai/TE-GCN)

Toolkit for UAV-Human Pose Estimation for TE-GCN

Generating splited UAV-Human Pose datasets (XSUB1 and XSUB2) for TE-GCN

More information about HAV_Human, plese refer HUA-Human

Extract the downloaded data Skeleton.zip into data folder.

Expected directory in data folder.

└───Skeleton
    ├───P000S00G10B10H10UC022000LC021000A000R0_08241716.txt
    ├───P000S00G10B10H10UC022000LC021000A001R0_08241716.txt

Run the following scripts to split the files into the train and the test set respectively:

python split_v1.py 
python split_v2.py 

Run the following script to preprocess the files:

python generate_data.py --data_path data/v1 
python generate_data.py --data_path data/v2 

Expected directory in data/v1

└───data/v1
    ├───train
        ├───P000S00G10B10H10UC022000LC021000A000R0_08241716.txt
        ├───P000S00G10B10H10UC022000LC021000A001R0_08241716.txt
        └───...
    ├───test
        ├───P000S00G10B10H10UC022000LC021000A000R0_08241716.txt
        ├───P000S00G10B10H10UC022000LC021000A001R0_08241716.txt
        └───...
    ├───train_label.pkl
    ├───train_data.npy
    ├───test_data.npy
    └───test_label.pkl

data/v1 has the similar structure