Ocean Pollution is one of the alarming environmental concerns where studies reveal that the biggest reason for ocean pollution is caused by the plastic debris discarded from the land. These plastics pose a threat to the coastal wildlife, marine ecosystem balance, and the economic health of the coastal communities. Inevitably this would result in affecting both human and aquatic living. The most commonly used methods, though effective, pose certain disadvantages when it comes to detecting and quantifying plastics. Thus, it is important to adopt alternative methods involving the latest technologies that would easily help us to identify the plastics and aid in their removal. In this paper, we have investigated the YOLO v4 and YOLO v5 deep learning object detection algorithms for detecting and identifying the marine plastics in the epipelagic layers of the water bodies. Ocean plastic images available on the internet are used to create the datasets. Image augmentation helps in increasing the number of images in the dataset. The Mean Average Precision of YOLO v4 and YOLO v5 are studied and the algorithm performance is explained with the results concluded.