Object Detection with OpenCV

My motivation behind this project was to control my computer by simple hand gestures through a webcam, without the need for external hardware.

I decided to approach this problem with the help of the OpenCV library.
After a series of object, skin and edge detection experiments, OpenCV proved as not the most robust way of accomplishing these computer vision tasks. Normally, this framework relies on specific color ranges as the basis for object detection. This is however highly sensitive to external circumstances, which led me to tackle these tasks from bottom-up and build my own machine learning pipeline.

For technical details and demos, please see my Github repo.

Handcrafted in New York City, design & code by
Dora Jambor