This repository provides training and evaluation code for paper titled "Graph Convolutional Networks for Assessment of Physical Rehabilitation Exercises" (accepted in IEEE TNSRE)
Dear fokhruli:
Thanks for your code and it really was an excellant project. But when I divide the dataset Kimore_ex5\TrainX.csv differently and train it with model ST-GCN, I get different results. I think it beacause of lack of data. For example, There are only 4 motions labeled 12.66666666.if I divide them all to the test set ,it will get a bad train result. I don't know if I am right,so I am looking forword to your reply. Have a good day!
Dear fokhruli: Thanks for your code and it really was an excellant project. But when I divide the dataset Kimore_ex5\TrainX.csv differently and train it with model ST-GCN, I get different results. I think it beacause of lack of data. For example, There are only 4 motions labeled 12.66666666.if I divide them all to the test set ,it will get a bad train result. I don't know if I am right,so I am looking forword to your reply. Have a good day!