Overview
VRL works with multiple groups of marine science researchers to develop novel computer vision solutions for their underwater videos and images. For example, the Marine Applied Research and Exploration (MARE) group collects videos of underwater habitats using an unmanned, remote controlled vehicle. They have spent thousands of human hours annotating and generating labels for substrates, habitats, and species occurring in these videos. We aim to help them automatically label future videos using existing labels as training data. Our machine learning methods can ultimately save their domain experts a lot of time, thereby saving them money and saving them from the tedious task of reviewing and labelling data.
