Gesture recognition proof of concept

Image processing pipeline: video > background removal (optional) > descale > gray scale conversion > fft > nearest neighbor classification

go build

Select the video interface

./eight -device <device>

There can be more than one point learned for a label.

./eight -learn <label>

To recognize gestures:

./eight -infer

To test the video interface:

./eight -picture

To use segmentation to remove background from frames:

./eight -segmentation

(doesn't work that well)