🎮 Tekken Gesture Control (Punch + Movement via Webcam)

Control your Tekken fighter using real-time upper body gestures captured through your webcam!
This project detects punches and body tilts using pose estimation powered by YOLOv8 Pose, triggering keypresses to control the game.

---

📦 Features

• 🥊 Punch Detection: Left and right punches using elbow-wrist-shoulder angles.
• ↔️ Tilt Detection: Move left or right based on upper body tilt.
• 🔁 Real-time Detection using YOLOv8 and OpenCV.
• ⌨️ Virtual Keypresses sent using the keyboard module.

---

📽️ Demo

Coming soon!
A short clip will be added here showing punches and movements mapped to keyboard keys.

---

🚀 Getting Started

1. Clone the Repository

git clone https://github.com/yourusername/tekken-gesture-control.git
cd tekken-gesture-control

2. Install Dependencies

pip install -r requirements.txt --extra-index-url https://download.pytorch.org/whl/cu118

⚠️ For GPU support, make sure PyTorch is installed with CUDA.

3. Download the YOLOv8 Pose Model

mkdir models
wget https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8n-pose.pt -P models/

Or place yolov8n-pose.pt manually inside the models/ directory.

🛠 Usage

Run the program in debug setup mode first to visualize detection:

python main.py

• Press q to exit setup mode.
• To disable debug overlay and begin controlling with gestures, set setup=False in main.py or modify the logic accordingly.

⚙️ Customization

You can tweak the following settings in config.py:

• KEY_MAP: Modify movement and attack keys
• Angle thresholds for punch detection
• Device preference (CPU/GPU)

🙌 Acknowledgements

• Ultralytics YOLOv8
• OpenCV
• PyTorch
• You — for trying this out!