Object detection in inland vessels using combined trained and pretrained models of YOLO8

Abstract —One of the main challenges in computer vision is object detection, which entails both locating and identifying specific items on an image. With a fresh perspective, the YOLO (You Only Look Once) algorithm was developed in 2015 and performs object detection in a single neural network. That caused the field of object detection to explode and produce considerably more amazing achievements than it had a decade before. So far, YOLO has been improved to eight versions and rated as one of the top object identification algorithms. This is thanks to its combination with many of the most cutting-edge concepts being explored in the computer vision research field. The most recent version of YOLO, known as YOLOv8, performs better than the YOLOv7 and YOLO5 in terms of accuracy and speed, though. This paper examines the most recent developments in computer vision that were incorporated into YOLOv5,YOLO7 and YOLO8 and its predecessors. Index Terms —Object Detection, YOLO, Autonomous Vehicles, Inland Waterway Vessels, Bounded Boxes, Neural Network, CNN. Received: 14 June 2023 Accepted: 11 September 2023 Published: 20 November 2023