The world beneath the ocean’s surface has always mesmerized the human mind and there is still so much left to explore. However, in recent times, the ocean has been polluted beyond repair by human activity, making it essential to take responsibility and work towards rectifying these mistakes. The accumulation of marine debris underwater affects the marine ecosystem and endangered aquatic animals. Hence, addressing this issue is essential for the safeguarding of the aquatic ecosystem and preserving the beauty of the ocean. While the process of addressing the issue of floating debris is straightforward, assessing the submerged marine debris is quite challenging due to the depths and unknown visibility of trash hotspots. Traditional methods of manual inspection and cleaning are labour intensive and time-consuming while this study explores advanced image processing techniques such as You Only Look Once and Real-Time Detection Transformer. The models are known for its real-time object detection, which is trained on a diverse dataset of underwater waste in various environmental conditions. This study also uses the enhancement of the underwater images using Dark Channel Prior for image dehazing. All the models are compared with each other before and after the image enhancements to ensure maximum efficiency and to validate its effectiveness. These models can then be used as algorithms for Automated Underwater Vehicles which are controlled remotely to clean the submerged marine debris.