Overview of Marine Object Detection Based on Image

Underwater object detection and recognition play a crucial role in marine scientific research and marine development. The optical features of underwater images are attenuated due to light scattering in the marine environment. Developing a model that satisfies the demands for detection precision and real-time performance while operating within the computational constraints of underwater robots poses a significant challenge. The intersection of progress in computing, underwater robotics, and artificial intelligence has garnered substantial scholarly attention in addressing this intricate issue. Underwater object recognition based on optical images or videos, under both artificial and natural lighting conditions, is aimed at achieving accurate detection and identification of submerged entities. This article succinctly delineates the procedural workflow for underwater object recognition, conducts a comprehensive scrutiny of cutting-edge algorithms, expounds upon the current research status and focal points in underwater recognition algorithmic research, and forecasts the prospective technological trajectory in this domain.