Hand Gesture Recognition (ML) Model for MysticGesturesGame

Dependencies

• Python >= 3.9.6
• OpenCV >= 4.11.0.86
• PyTorch >= 2.6.0
• MediaPipe >= 0.10.21
• Pandas >= 2.2.3
• Scikit-learn == 1.5.1

Usage

Argument	Type	Default Value	Description
--device	int	0	Specifies the device to use (e.g., camera ID).
--width	int	960	Sets the capture width.
--height	int	540	Sets the capture height.
--use_static_image_mode	bool	False	Enables or disables static image mode.
--min_detection_confidence	float	0.7	The minimum confidence threshold for detection.
--min_tracking_confidence	float	0.5	The minimum confidence threshold for tracking.
--server_address	str	'127.0.0.1'	The UDP server address to connect to.
--server_port	int	65432	The UDP server port to connect to.
--stability_threshold	int	2000	Sets the stability threshold in milliseconds for prediction.

# Example usage (UV)
$ uv run python model/main.py --device=1 --width=1280 --height=720 --server_address='192.168.1.100' --server_port=12345

Output

ID	Label	Gesture
0	moves	Move
1	fire	Fire
2	explode	Explode
3	wind	Wind
4	shrink	Shrink

License & Copyright

This repository is a modification of hand-gesture-recognition-using-mediapipe (Takahashi, 2020) which is under the Apache v2 license.

Modification Summary

• TODO: Add modification summaries
• Added hand gesture stability check
• Added UDP socket message sending

References

Takahashi, S. (2020) Hand Gesture Recognition Using MediaPipe. Available at: https://github.com/Kazuhito00/hand-gesture-recognition-using-mediapipe (Accessed: 5 April 2025).