pp. 3441-3450
S&M3403 Research Paper of Special Issue https://doi.org/10.18494/SAM4477 Published: September 29, 2023 Marine Debris Detection Using Optimized You Only Look Once Version 5 [PDF] Tae-Young Lee, Seung Bae Jeon, and Myeong-Hun Jeong (Received April 27, 2023; Accepted August 8, 2023) Keywords: marine debris, object detection, marine observation, deep learning, YOLOv5
Marine debris is one of the most widespread pollution problems facing oceans and waterways. It threatens the marine ecosystem and navigation safety. Thus, the extensive and timely monitoring of marine debris is crucial. In this study, we aimed to detect marine debris from images. You Only Look Once Version 5 (YOLOv5) was used to train a model to detect marine debris. Our experimental results reveal that the model achieves a mean average precision (mAP) of 96.8% and an inference speed of 1.3 ms per image. The optimized YOLOv5 has a higher mAP than a faster region-based convolutional neural network and requires less inference time than a single-shot multibox detector.
Corresponding author: Myeong-Hun Jeong This work is licensed under a Creative Commons Attribution 4.0 International License. Cite this article Tae-Young Lee, Seung Bae Jeon, and Myeong-Hun Jeong , Marine Debris Detection Using Optimized You Only Look Once Version 5, Sens. Mater., Vol. 35, No. 9, 2023, p. 3441-3450. |