Citation: Wang, J.; Qi, S.; Wang, C.;
Luo, J.; Wen, X.; Cao, R. B-YOLOX-S:
A Lightweight Method for
Underwater Object Detection Based
on Data Augmentation and
Multiscale Feature Fusion. J. Mar. Sci.
Eng. 2022, 10, 1764. https://doi.org/
10.3390/jmse10111764
Academic Editors: Simone Marini,
Jacopo Aguzzi, Giacomo Picardi,
Damianos Chatzievangelou, Sascha
Flögel, Sergio Stefanni, Peter Weiss
and Daniel Mihai Toma
Received: 8 October 2022
Accepted: 11 November 2022
Published: 16 November 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Journal of
Marine Science
and Engineering
Article
B-YOLOX-S: A Lightweight Method for Underwater
Object Detection Based on Data Augmentation and Multiscale
Feature Fusion
Jun Wang, Shuman Qi, Chao Wang, Jin Luo, Xin Wen and Rui Cao *
College of Software, Taiyuan University of Technology, Taiyuan 030024, China
* Correspondence: caorui@tyut.edu.cn; Tel.: +86-132-3368-1616
Abstract:
With the increasing maturity of underwater agents-related technologies, underwater object
recognition algorithms based on underwater robots have become a current hotspot for academic and
applied research. However, the existing underwater imaging conditions are poor, the images are
blurry, and the underwater robot visual jitter and other factors lead to lower recognition precision and
inaccurate positioning in underwater target detection. A YOLOX-based underwater object detection
model, B-YOLOX-S, is proposed to detect marine organisms such as echinus, holothurians, starfish,
and scallops. First, Poisson fusion is used for data amplification at the input to balance the number of
detected targets. Then, wavelet transform is used to perform Style Transfer on the enhanced images
to achieve image restoration. The clarity of the images and detection targets is further increased
and the generalization of the model is enhanced. Second, a combination of BIFPN-S and FPN is
proposed to fuse the effective feature layer obtained by the Backbone layer to enhance the detection
precision and accelerate model detection. Finally, the localization loss function of the prediction
layer in the network is replaced by EIoU_Loss to heighten the localization precision in detection.
Experimental results comparing the B-YOLOX-S algorithm model with mainstream algorithms such
as FasterRCNN, YOLOV3, YOLOV4, YOLOV5, and YOLOX on the URPC2020 dataset show that the
detection precision and detection speed of the algorithm model have obvious advantages over other
algorithm networks. The average detection accuracy mAP value is 82.69%, which is 5.05% higher
than the benchmark model (YOLOX-s), and the recall rate is 8.03% higher. Thus, the validity of the
algorithmic model proposed in this paper is demonstrated.
Keywords: object detection; YOLOX; data augmentation; URPC
1. Introduction
With the vigorous growth of target detection in computer vision, underwater tar-
get detection based on optical imaging plays an important role in fishery, aquaculture,
underwater archaeology, marine military, and other fields [
1
–
4
]. In the field of marine
fishery, traditional underwater frogmen fish and explore; they require a lot of equipment
support and sufficient underwater experience, and they are also faced with life-threatening
situations at any time. Long-term fishing operations lead to serious occupational diseases,
and the cost of manual fishing operations is gradually increasing [
5
]. Due to the limitation
of fishing time and the impact of the marine environment, fishing operations provide great
challenges. For example, the habitat of seafood is in the bottom of rocky reefs in the deep
sea and in sediment with dense water and grass [
6
]. It is a key task to adopt underwater
object detection network algorithms to improve fishing accuracy.
Nowadays, with the rapid progress of Deep learning and its excellent performance in
various fields, an increasing number of scholars are working on the use of deep learning
methods in underwater target detection [
7
]. However, due to the limitations of the marine
environment, large-scale fishing equipment cannot be used in marine pastures. Nowadays,
fishing operations are mainly carried out using underwater robots. Because of the limitation
J. Mar. Sci. Eng. 2022, 10, 1764. https://doi.org/10.3390/jmse10111764 https://www.mdpi.com/journal/jmse
