Hand Gesture Controlled Drone

This is a hand gesture based drone control project built using MediaPipe and CNN.

The system detects hand gestures from a live camera feed and converts them into drone control commands.
A dummy drone controller is used to simulate drone movement.

---

Gestures Supported

• UP – Move drone up
• DOWN – Move drone down
• LEFT – Move drone left
• RIGHT – Move drone right
• CIRCLE (V sign) – Drone circles around the person

If no gesture is detected, the drone holds its position.

---

How It Works

Camera → MediaPipe Hand Landmarks → CNN / Rule Logic → Drone Commands

MediaPipe detects 21 hand landmarks (63 values).
These values are used to detect gestures either:

• directly using rules (MediaPipe only), or
• using a CNN trained on a custom dataset.

---

Files

• collect_data.py – Collect gesture data
• train_cnn.py – Train CNN model
• live_cnn_gesture.py – Live gesture detection (CNN)
• mediapipe_only_control.py – Gesture detection using only MediaPipe
• drone_controller.py – Dummy drone control
• circle_motion.py – Circling logic

---

How to Run

Activate virtual environment: Run the live_cnn_gesture.py file