Search and Detection of People in the Water Using YOLO Architectures: a Comparative Analysis from YOLOv3 to YOLOv8

EasyChair Preprint 15000

Abstract

The rapid development of computer vision and deep learning technologies has significantly improved the accuracy and speed of object recognition in a variety of applications, including security, surveillance, and search and rescue. One of the key challenges in this area is the detection of people in water areas, which is crucial for improving water safety and emergency response. In this research, a detailed comparative analysis of YOLOv3 to YOLOv8, is performed to evaluate their ability for effective detection people in the water. The analysis focuses on assessing the accuracy of each version's identification of people in the water, the speed of real-time image processing, the ability to adapt to different water conditions, and the required computing resources for effective operation. The purpose of the research is to perform a detailed comparative analysis of YOLO architectures, from YOLOv3 to YOLOv8, for evaluating their ability to effectively detect people in the water area. The research not only assesses current capabilities, but also suggests directions for future innovation to improve the efficiency and reliability of detecting and tracking people on water.

Keyphrases: YOLO, deep learning, object detection

@booklet{EasyChair:15000,
  author    = {Nataliya Bilous and Vladyslav Malko and Nazarii Moshenskyi},
  title     = {Search and Detection of People in the Water Using YOLO Architectures: a Comparative Analysis from YOLOv3 to YOLOv8},
  howpublished = {EasyChair Preprint 15000},
  year      = {EasyChair, 2024}}