Research Article
Design and Evaluation of Efficient Underwater Object Detection Systems Using AI
@INPROCEEDINGS{10.4108/eai.28-4-2025.2358015, author={Seelam Ruchitha Mahi and Siddamsetti Vamsi Krishna and Annam Komal Sai Mani Naga Vaibhav and Prayaga Vathsalya Sri Vijaya and Brij Kishor Tiwari}, title={Design and Evaluation of Efficient Underwater Object Detection Systems Using AI}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part II}, publisher={EAI}, proceedings_a={ICITSM PART II}, year={2025}, month={10}, keywords={underwater object detection deep learning yolov8 yolov10 yolov11 yolov12 mean average precision (map)}, doi={10.4108/eai.28-4-2025.2358015} }- Seelam Ruchitha Mahi
Siddamsetti Vamsi Krishna
Annam Komal Sai Mani Naga Vaibhav
Prayaga Vathsalya Sri Vijaya
Brij Kishor Tiwari
Year: 2025
Design and Evaluation of Efficient Underwater Object Detection Systems Using AI
ICITSM PART II
EAI
DOI: 10.4108/eai.28-4-2025.2358015
Abstract
Underwater object detection plays a crucial role in marine applications, where accurate detection is often prevented by obstacles like poor visibility, image distortion, and environmental noise. To address these issues, this study examines the effectiveness of the various deep learning models in detecting underwater objects. In this study YOLOv8, YOLOv10, YOLOv11, and YOLOv12 models were used. These model’s performance on a wide range of underwater images is assessed based on their accuracy (mean average precision, or mAP), speed, and capacity to operate in challenging circumstances. Each model was trained and tested using the same data to provide a fair comparison. The results indicate that the more recent YOLO models, particularly YOLOv11 and YOLOv12, achieve high accuracy and speed marks, with an average mean precision of 91.5%. This study proves that using the most recent YOLO models can improve underwater object detection and support real-time marine applications like underwater exploration and autonomous underwater vehicles.
