MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 model
Underwater light attenuation leads to decreased image contrast. This reduction in contrast subsequently decreases target visibility. Additionally, marine target detection is challenging due to multi-scale problems from varying target-to-device distances, complex target clustering, and noise from wat...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2024-02-01
|
| Series: | Heliyon |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2405844024021765 |
| _version_ | 1827322230480044032 |
|---|---|
| author | W.E.I. Lian-suo Huang Shen-hao Ma Long-yu |
| author_facet | W.E.I. Lian-suo Huang Shen-hao Ma Long-yu |
| author_sort | W.E.I. Lian-suo |
| collection | DOAJ |
| description | Underwater light attenuation leads to decreased image contrast. This reduction in contrast subsequently decreases target visibility. Additionally, marine target detection is challenging due to multi-scale problems from varying target-to-device distances, complex target clustering, and noise from waterborne particulates.To address these issues, we propose MTD-YOLOv5.Initially, we enhance image contrast with grayscale equalization and mitigate color shift issues through color space transformation.We then introduce a novel feature extraction module, PCBR, combining max pooling and convolution layers for more effective target feature extraction from the background.Furthermore, we present the Multi-Scale Perceptual Hybrid Pooling (MHP) module.This module integrates horizontal and vertical receptive fields to establish long-range dependencies, thereby capturing hidden target information in deep network feature maps. In the Labeled Fishes in the Wild test datasets, MTD-YOLOv5 achieves a precision of 88.1% and a mean Average Precision (mAP[0.5:.95]) of 49.6%.These results represent improvements of 2.6% in precision and 0.4% in mAP over the original YOLOv5. |
| first_indexed | 2024-03-08T00:09:02Z |
| format | Article |
| id | doaj.art-5087e9569eb84baf94dafc25df498cb3 |
| institution | Directory Open Access Journal |
| issn | 2405-8440 |
| language | English |
| last_indexed | 2024-04-25T01:20:12Z |
| publishDate | 2024-02-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Heliyon |
| spelling | doaj.art-5087e9569eb84baf94dafc25df498cb32024-03-09T09:27:22ZengElsevierHeliyon2405-84402024-02-01104e26145MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 modelW.E.I. Lian-suo0Huang Shen-hao1Ma Long-yu2School of information engineering, Suqian University, SuQian, jiangsu 223800, China; Corresponding author.College of Computer & Control Engineering, Qiqihar University, Qiqihar, Heilongjiang, 161006, ChinaCollege of Computer & Control Engineering, Qiqihar University, Qiqihar, Heilongjiang, 161006, ChinaUnderwater light attenuation leads to decreased image contrast. This reduction in contrast subsequently decreases target visibility. Additionally, marine target detection is challenging due to multi-scale problems from varying target-to-device distances, complex target clustering, and noise from waterborne particulates.To address these issues, we propose MTD-YOLOv5.Initially, we enhance image contrast with grayscale equalization and mitigate color shift issues through color space transformation.We then introduce a novel feature extraction module, PCBR, combining max pooling and convolution layers for more effective target feature extraction from the background.Furthermore, we present the Multi-Scale Perceptual Hybrid Pooling (MHP) module.This module integrates horizontal and vertical receptive fields to establish long-range dependencies, thereby capturing hidden target information in deep network feature maps. In the Labeled Fishes in the Wild test datasets, MTD-YOLOv5 achieves a precision of 88.1% and a mean Average Precision (mAP[0.5:.95]) of 49.6%.These results represent improvements of 2.6% in precision and 0.4% in mAP over the original YOLOv5.http://www.sciencedirect.com/science/article/pii/S2405844024021765YOLOv5Marine target detectionFeature enhancementMulti scale perceptual field |
| spellingShingle | W.E.I. Lian-suo Huang Shen-hao Ma Long-yu MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 model Heliyon YOLOv5 Marine target detection Feature enhancement Multi scale perceptual field |
| title | MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 model |
| title_full | MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 model |
| title_fullStr | MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 model |
| title_full_unstemmed | MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 model |
| title_short | MTD-YOLOv5: Enhancing marine target detection with multi-scale feature fusion in YOLOv5 model |
| title_sort | mtd yolov5 enhancing marine target detection with multi scale feature fusion in yolov5 model |
| topic | YOLOv5 Marine target detection Feature enhancement Multi scale perceptual field |
| url | http://www.sciencedirect.com/science/article/pii/S2405844024021765 |
| work_keys_str_mv | AT weiliansuo mtdyolov5enhancingmarinetargetdetectionwithmultiscalefeaturefusioninyolov5model AT huangshenhao mtdyolov5enhancingmarinetargetdetectionwithmultiscalefeaturefusioninyolov5model AT malongyu mtdyolov5enhancingmarinetargetdetectionwithmultiscalefeaturefusioninyolov5model |