DeepSea: Underwater Marine Species Detection

My CS230 team developed a computer vision-based marine species detection system optimized for turbid underwater environments in order to automate the laborious cataloging process common in marine research and to complement underwater robotics used for deep sea exploration. Our approach involved augmenting a dataset of aquarium images with synthetic ocean-like noise. We then coupled a fine-tuned YOLOv5su model with our custom MotionAwareDenoiser to classify marine species in images.