[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 

Saved Queries

Sign in to use this feature.

Search Filter Reset All

Years

Between: -

Subjects

remove_circle_outline
remove_circle_outline
remove_circle_outline
Add Subjects Reset
Select Subjects

Journals

remove_circle_outline
Add Journals Reset
Select Journals

Article Types

remove_circle_outline
Add Article Types Reset
Select Article Types

Countries / Regions

remove_circle_outline
Add Countries / Regions Reset
Select Countries / Regions
Update Search
share announcement

Search Results (1)

Search Parameters:
Keywords = BGLE-YOLO

Order results
Result details
Results per page
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
22 pages, 6298 KiB  
Article
BGLE-YOLO: A Lightweight Model for Underwater Bio-Detection
by Hua Zhao, Chao Xu, Jiaxing Chen, Zhexian Zhang and Xiang Wang
Sensors 2025, 25(5), 1595; https://doi.org/10.3390/s25051595 - 5 Mar 2025
Viewed by 129
Abstract
Due to low contrast, chromatic aberration, and generally small objects in underwater environments, a new underwater fish detection model, BGLE-YOLO, is proposed to investigate automated methods dedicated to accurately detecting underwater objects in images. The model has small parameters and low computational effort [...] Read more.
Due to low contrast, chromatic aberration, and generally small objects in underwater environments, a new underwater fish detection model, BGLE-YOLO, is proposed to investigate automated methods dedicated to accurately detecting underwater objects in images. The model has small parameters and low computational effort and is suitable for edge devices. First, an efficient multi-scale convolutional EMC module is introduced to enhance the backbone network and capture the dynamic changes in targets in the underwater environment. Secondly, a global and local feature fusion module for small targets (BIG) is integrated into the neck network to preserve more feature information, reduce error information in higher-level features, and increase the model’s effectiveness in detecting small targets. Finally, to prevent the detection accuracy impact due to excessive lightweighting, the lightweight shared head (LSH) is constructed. The reparameterization technique further improves detection accuracy without additional parameters and computational cost. Experimental results of BGLE-YOLO on the underwater datasets DUO (Detection Underwater Objects) and RUOD (Real-World Underwater Object Detection) show that the model achieves the same accuracy as the benchmark model with an ultra-low computational cost of 6.2 GFLOPs and an ultra-low model parameter of 1.6 MB. Full article
(This article belongs to the Topic Applications and Development of Underwater Robotics and Underwater Vision Technology)
►▼ Show Figures

Figure 1

Figure 1
<p>Images (<b>a</b>,<b>b</b>) describe underwater image features marked by low contrast and small targets, respectively, and (<b>c</b>,<b>d</b>) describe underwater image features marked by underwater blurring and color deviations due to various attenuations, respectively.</p> Full article ">Figure 2
<p>(<b>a</b>) The diagram illustrates the architecture of YOLOv8. (<b>b</b>) The diagram depicts the architecture of BGLE-YOLO. Compared to (<b>a</b>), (<b>b</b>) adds BiFPN network, GLSA attention block, EMC convolution, and LSH detection header to (<b>a</b>).</p> Full article ">Figure 3
<p>The structures of EMC are presented. The input feature map is channelized and then fused into an output feature map by independent multi-channel features.</p> Full article ">Figure 4
<p>The structures of FPN, PANet, NAS-FPN, and BiFPN.</p> Full article ">Figure 5
<p>Introduction to the Global-to-Local Spatial Aggregation (GLSA) module.</p> Full article ">Figure 6
<p>Convolution in LSH consists of the group normalized GN convolution and the group normalized detail-enhanced convolution DEConv. The red pixels are normalized using the same mean and variance, which are calculated by combining the values of these pixels.</p> Full article ">Figure 7
<p>Comparison plots of the YOLO family of algorithms on the DUO dataset. (<b>a</b>) Accuracy comparison plot; (<b>b</b>) mAP@0.5 comparison plot; (<b>c</b>) mAP@0.5:0.95 comparison plot.</p> Full article ">Figure 8
<p>Comparison plots of the YOLO family of algorithms on the RUOD dataset. (<b>a</b>) Accuracy comparison plot; (<b>b</b>) mAP@0.5 comparison plot; (<b>c</b>) mAP@0.5:0.95 comparison plot.</p> Full article ">Figure 9
<p>Comparison of parameters as well as computational effort of different models.</p> Full article ">Figure 10
<p>Qualitative comparison of the detection performance of the YOLO series of models, (<b>a</b>–<b>c</b>) showing the detection results for four categories in the DUO dataset.</p> Full article ">Figure 11
<p>Qualitative comparison of the detection performance of the YOLO series of models, (<b>a</b>–<b>c</b>) showing the detection results for four categories in the RUOD dataset.</p> Full article ">
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying article 1-50 on page 1 of 1.
Back to TopTop