Anomaly Detection In Auv Images
Classifying and categorizing anomolous data gathered from a deep sea autonomous underwater vehicle (AUV).
This was a project completed Spring 2020 during the Machine Learning course at Georgia Tech. Our team acquired data from the IceFin AUV composed of 2,000 images of the Antartic seabed. We trained a CNN to classify images as either anomolous or non-anomolous. From this classified data, we used unsupervised machine learning algorithms to group similar images. See the full project as well as Github repo below.