Too Long; Didn't Read
DeepPlastic is a novel approach that uses Deep Learning to identify marine-plastic. Using our model, we can now detect on average 85% (mAP) of all epipelagic plastic in the ocean. We achieve this level of precision based on a Neural Network architecture called YOLOv5-S. The most common monitoring method to quantify floating plastic requires the use of a manta trawl. Without better monitoring and sampling methods, the total impact of plastic pollution on the environment as a whole, and details of impact within specific oceanic regions, will remain unknown.