Lightweight Deep Neural Network for Joint Learning of Underwater Object Detection and Color Conversion

Abstract—Underwater image processing has been shown to exhibit significant potential for exploring underwater environ- ments. It has been applied to a wide variety of fields, such as underwater terrain scanning and autonomous underwater vehi- cles (AUVs)-driven applications, such as image-based underwater object detection. However, underwater images often suffer from degeneration due to attenuation, color distortion, and noise from artificial lighting sources as well as the effects of possibly low- end optical imaging devices. Thus, object detection performance would be degraded accordingly. To tackle this problem, in this article, a lightweight deep underwater object detection network is proposed. The key is to present a deep model for jointly learning color conversion and object detection for underwater images. The image color conversion module aims at transforming color images to the corresponding grayscale images to solve the problem of underwater color absorption to enhance the object detection performance with lower computational complexity. The presented experimental results with our implementation on the Raspberry pi platform have justified the effectiveness of the proposed lightweight jointly learning model for underwater object detection compared with the state-of-the-art approaches. Index Terms—Convolutional neural networks, deep learning, lightweight deep model, underwater image processing, underwa- ter object detection. Manuscript received 13 March 2020; revised 1 October 2020 and 21 January 2021; accepted 30 March 2021. Date of publication 26 April 2021; date of current version 28 October 2022. This work was supported in part by the Ministry of Science and Technology, Taiwan, under Grant NSC 102-2221-E-110-032-MY3, Grant MOST 103-2221-E-110-045-MY3, Grant MOST 103-2221-E-003-034-MY3, Grant MOST 105-2221-E-003-030-MY3, Grant MOST 108-2221-E-003-027-MY3, Grant MOST 108-2218-E-003- 002, Grant MOST 108-2218-E-110-002, Grant MOST 109-2218-E-110-007, Grant MOST 109-2224-E-110-001, and Grant MOST 109-2218-E-003-002; and in part by the Intelligent Recognition Industry Service Center, National Yunlin University of Science and Technology, Douliu, through the Fea- tured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan. (Corresponding author: Li-Wei Kang.) Chia-Hung Yeh is with the Department of Electrical Engineering, National Taiwan Normal University, Taipei 106, Taiwan, and also with the Department of Electrical Engineering, National Sun Yat-sen University, Kaohsiung City 80021, Taiwan. Chu-Han Lin, Chih-Hsiang Huang, Min-Hui Lin, and Chua-Chin Wang are with the Department of Electrical Engineering, National Sun Yat-sen University, Kaohsiung City 80021, Taiwan. Li-Wei Kang is with the Department of Electrical Engineering, National Taiwan Normal University, Taipei 106, Taiwan (e-mail: lwkang@ntnu.edu.tw). Chuan-Yu Chang is with the Department of Computer Science and Infor- mation Engineering, National Yunlin University of Science and Technology, Douliu 64002, Taiwan. Color versions of one or more figures in this article are available at https://doi.org/10.1109/TNNLS.2021.3072414. Digital Object Identifier 10.1109/TNNLS.2021.3072414