One-stage CNN detector-based benthonic organisms detection with limited training dataset

Tingkai Chen; Ning Wang; Rongfeng Wang; Hong Zhao; Guichen Zhang

doi:10.1016/j.neunet.2021.08.014

One-stage CNN detector-based benthonic organisms detection with limited training dataset

Neural Netw. 2021 Dec:144:247-259. doi: 10.1016/j.neunet.2021.08.014. Epub 2021 Aug 28.

Authors

Tingkai Chen¹, Ning Wang², Rongfeng Wang³, Hong Zhao⁴, Guichen Zhang⁵

Affiliations

¹ School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China. Electronic address: tingkai.chen@hotmail.com.
² School of Marine Engineering, Dalian Maritime University, Dalian 116026, China; College of Shipbuilding Engineering, Harbin Engineering University, Harbin 150001, China. Electronic address: n.wang@ieee.org.
³ School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China. Electronic address: rfwang.ee@foxmail.com.
⁴ School of Marine Electrical Engineering, Dalian Maritime University, Dalian 116026, China. Electronic address: zhaohong@dlmu.edu.cn.
⁵ Merchant Marine College, Shanghai Maritime University, Shanghai 201306, China. Electronic address: gczhang@shmtu.edu.cn.

PMID: 34508919
DOI: 10.1016/j.neunet.2021.08.014

Abstract

In this paper, focusing on the challenges in unique shape dimension and limited training dataset of benthonic organisms, an one-stage CNN detector-based benthonic organisms detection (OSCD-BOD) scheme is proposed. Main contributions are as follows: (1) The regression loss between the predicted bounding box and ground truth box is innovatively measured by the generalized intersection over union (GIoU), such that localization accuracy of benthonic organisms is dramatically enhanced. (2) By devising K-means-based dimension clustering, multiple benthonic organisms anchor boxes (BOAB) sufficiently exploring a priori dimension information can be finely derived from limited training dataset, and thereby significantly promoting the recall ability. (3) Geometric and color transformations (GCT)-based data augmentation technique is further resorted to not only efficiently prevent over-fitting training but also to significantly enhance detection generalization in complex and changeable underwater environments. (4) The OSCD-BOD scheme is eventually established in a modular manner by integrating GIoU, BOAB and GCT functionals. Comprehensive experiments and comparisons sufficiently demonstrate that the proposed OSCD-BOD scheme outperforms typical approaches including Faster R-CNN, SSD, YOLOv2, YOLOv3 and CenterNet in terms of mean average precision by 6.88%, 10.92%, 12.44%, 3.05% and 1.09%, respectively.

Keywords: Benthonic organisms anchor boxes; Benthonic organisms detection; Data augmentation; Generalized intersection over union; One-stage CNN detector.