It's our Computer Graphics final project. There are 5 applications of gesture-operation.
Some of them are referenced in the cvzone tutorial. They are all based on mediapipe and camera of computer. Here is final video.
All codes are based on x86 Python. Run
pip install requirements.txt
to install all dependencies.
Our detection part is made in use of mediapipe
, a frame work to dect human working of google.
Its data is like this.(the data: hand).
[
{'lmList': [
[478, 523, 0], #[x, y, z]
...
[525, 294, -15],
],
'bbox': (478, 275, 197, 248),
'center': (576, 399),
'type': 'Left'
}
]
Key point locations are stored as [x, y, z]
of Hand Landmark Model in lmList
. The sequence of key points is shown in the figure.
Here's an example.
See more details of demos in their own README.md
and ./report/report.pdf