Target Classification of Marine Debris Using Deep Learning

Marine Debris is human-created waste dumped into the sea or ocean. It pollutes aquatic environment and hence very dangerous for ocean species. Removal of marine debris from necessary to eliminate pollution secure life. A robust automatic system essential that detects unnecessary litter plastic other garbage at real-time. In this study, we have proposed deep learning based architecture detection classification debris. Histogram Equalization technique combined with Median Filter used enhance contrast images remove noise. Experiments are performed on challenging Forward Looking Sonar Image (FLS) Dataset. This dataset includes ten different types Debris. The not only detect Debris, but also classify it classes. To overcome challenge data scarcity, Faster-RCNN transfer ResNet-50 used. one popular object uses Regional Proposal Network (RPN) detector same time. methodology significantly improves state-of-the-art results. Result assessment our achieved recall (96%) Mean Overlap bounding boxes (3.78). Visual qualitative shows effectiveness presented technique.