Jiangtao-ACCV 2020-LU version.pdf (2.59 MB)
Download fileCompact and fast underwater segmentation network for autonomous underwater vehicles
conference contribution
posted on 2021-12-20, 11:03 authored by Jiangtao WangJiangtao Wang, Baihua LiBaihua Li, Yang ZhouYang Zhou, Emanuele Rocco, Qinggang MengQinggang MengReliable and real-time semantic segmentation is crucial for vision-based navigation tasks undertaken by AUVs (Autonomous Underwater Vehicles). However state-of-art deep learning segmentation networks could not be deployed on embedded devices with limited onboard resources, due to the required high computation capacity and the lack of capability to deal with poor underwater image quality. In this work we present a new deep underwater segmentation network, featured by a compact encoder and a lightweight decoder. We use only one step upsampling block to recover features maps from the encoder to significantly speed up the inference time. Furthermore, we adopt three strategies to improve the network accuracy. Firstly, in parallel with the main decoder path, we introduce a branch path to extract additional low-level features. Secondly, we use position attention module to enhance the high-level semantic information and use channel attention module to introduce extra global context as well as refine the inter-dependencies of each features. Thirdly, we proposed to use two additional auxiliary loss and smooth loss functions to better train the network, such that it will be more robust in segmenting images at varying resolutions and generating smooth boundaries. We validate our network accuracy on two different underwater segmentation datasets, a generalistic and a specialist one, and our model achieves the same level of accuracy of state-of-art networks. We also tested the network speed on different embedded platforms, and we showed it reaches real-time inference speed on both Nvidia Jetson GPU platforms TX2 and Nano, with respectively around 24 and 18 FPS (Frame Per Second). The proposed network inference is up to 27 times faster than other considered networks. Its high accuracy and speed will so pave the way for its deployment and application on AUVs systems.
Funding
EPSRC Centre for Doctoral Training in Embedded Intelligence
Engineering and Physical Sciences Research Council
Find out more...History
School
- Science
Department
- Computer Science
Published in
Computer Vision – ACCV 2020: 15th Asian Conference on Computer Vision, Kyoto, Japan, November 30 – December 4, 2020, Revised Selected Papers, Part IIIPages
688 - 703Source
15th Asian Conf. on Computer Vision. ACCV’20Publisher
Springer International PublishingVersion
- AM (Accepted Manuscript)
Publisher statement
This book chapter was published in the book Computer Vision – ACCV 2020. The publisher's website is at https://doi.org/10.1007/978-3-030-69535-4_42Acceptance date
2020-09-17Publication date
2021-02-25Copyright date
2021ISBN
9783030695347ISSN
0302-9743eISSN
1611-3349Publisher version
Book series
Lecture Notes in Computer Science (LNCS, volume 12624)Language
- en