I select 6 different gestures from jester dataset
Include swiping left, swiping right, no gesture, rolling forward, rolling backward, stop
First model:LRN+HRN can reach 0.78 accuracies
Second model: 3DCNN can reach 0.76 accuracies
Third model: 3DCNN+ one layer LSTM / + two layer LSTM / + three layer LSTM
python real_time/real_time.py
I also post a video on youtube:
https://www.youtube.com/watch?v=vd0czNKL1DE
Feel free to contact me on linkdein:
https://www.linkedin.com/in/wayne-shang-909636166/