This code was used as a proof of concept for an experiment for the following abstract: https://cdn.ymaws.com/siim.org/resource/resmgr/siim2019/abstracts/ML_Technique_Goel.pdf
gen-us.py
samples images from Cine US images and combines these images with random strings. During the insertion of strings, the bounding boxes are also stored as labels for the data.
keras-model.py
utilizes the generated data to create a model that predicts the text location in the image.