a4aleem / flywave

Controlling a drone using deep learning and computer vision to detect arm gestures.

Project FlyWave

The goal of the FlyWave team of UAVs@Berkeley is to develop a drone that can maneuver autonomously using computer vision and deep learning to detect hand gestures.

Team

• Alex Chan (Project Lead)
• Timothy Liu
• Raymond Gu
• Nick Mecklenburg

Highlights

• Person classification
• Hand gesture detection
• Person movement tracking
• Gesture detection
• Using gestures to move the drone

Requirements

1. Parrot Bebop 2
2. Packages: Tensorflow, Keras, OpenCV, and PyParrot (dependencies described here).

Setup

1. Connect to the Bebop's Wifi.
2. Before running any scripts, stand clear of the drone's takeoff area.
3. In terminal, run:

python drone_movement.py

4. Stand in front of Bebop, and make gestures as desired. Right arm out: Move bebop in an arc to the right Left arm out: Move bebop in an arc to the left Right arm up: Land drone Left arm up: Flip

Credits

Credit to: PyParrot

Copyright for object detection: See LICENSE for details. Copyright (c) 2017 Dat Tran.