This was an incredibly fun project which I got to work on with some good friends, ultimately winning top prize at HackGT6! Essentially we were tasked with locating anomalies in seismic data. As you might imagine, the DoD was pretty vague about what these "anomalies" really were, so we just had to operate in the dark there. We used various filters and convolution to transform our data in a way to make the anomalies more clear. Then we used OpenCV to place bounding boxes around the anomalies.
Here's the full post!