Ocean water is important to all human beings. Marine life such as fish, corals, seaweeds, and seagrass are critical organisms for the ocean ecosystem. A robust automated system is increasingly needed to monitor the marine ecosystem, which has proved helpful for all researchers in their quest to gather data on marine species. In this paper, yolo-based algorithms for seaweeds detection are evaluated. Each training image was manually examined, and bounding boxes were created for every seaweeds found in the training image. Microsoft vott was used for the bounding boxes. A pre-trained weights was used for training the YOLOV3 model. Each sample of bounding boxes was normalized to (640,640). A total of 2000 samples were used to train the YOLOV3 model. The detection rate of the YOLOV3 is 73%. To perform further experiment, YOLOV5 was also examined to check for accuracy and performance improvement. The experiment showed that YOLOV5 is faster compared to YOLOV3 but produces lower accuracy result for 3 to 5% reduction as compared to YOLOV3.