davidflanagan / Lighthouse-2

Experiments with image processing to help blind people

More experiment for helping out blind people

For the moment, these experiments concentrate on detecting objects shaken in front of the camera. Later, we will try to support a moving camera, as well as storing the picture, matching new pictures against stored ones.

Basic usage

Launch:

python src/main.py

Move an object in front of the camera. It will try to isolate what is moving.

Image matching

To extract and save features from the image set you can use the following command:

$ python ./src/matching/extract_features.py -i ./samples/products-front-back -o ./features.json [--detector={orb, akaze, surf}] [--orb-n-features=2000] [--verbose]

To run image matching you can use the following command:

$ python ./src/matching/match.py -t ./samples/products-front-back/product-1-front.jpg -i ./samples/products-front-back [--detector={orb, akaze, surf}] [--orb-n-features=2000] [--ratio-test-k=0.75] [--n-matches=3] [--no-ui] [--verbose]

or if you already have a file with serialized image features:

$ python ./src/matching/match.py -t ./samples/products-front-back/product-1-front.jpg -d ./features.json [--detector={orb, akaze, surf}] [--orb-n-features=2000] [--ratio-test-k=0.75] [--n-matches=3] [--no-ui] [--verbose]

or run $ python ./src/matching/match.py -h to see all available options.