Abstract
Traditional tracking algorithms reply on manually extracted features while deep learning algorithms can automatically extract features, which is of great importance for single target tracking in the complex underwater environment that is characterized by poor visibility, low contrasts, and occlusion. The convolutional neural network SiamRPN++ is an advanced deep learning algorithm whose backbone network structure is. However, this algorithm is difficult to be implemented in an underwater platform due to its large amount of computation and high requirement for the computing power of the hardware. To solve this problem, a new backbone network structure NewNet-62 characterized by inverted residual bottleneck block and the SiamRPN++ algorithm are proposed. The depth-wise separable convolution in NewNet-62 simplifies the computation so that the SiamRPN++ algorithm can realize fast and accurate tracking of underwater single targets. The video set in VOT2016 was used to test the algorithm. The results show that the SiamRPN++ algorithm using NewNet-62 as the main network has the great performance. Compared with the algorithms using ResNet-50, the accuracy of the SiamRPN++ algorithm increased to 0.524 and its EAO increased to 0.303. And its tracking speed increased to 73.74. The network complexity reduced to 3.013 billion and the network parameters reduced to 12.538 million, which significantly reduces network complexity and number of network parameters.
Highlights
In order to cope with the depletion of land resources and the deterioration of the ecological environment, people have turned their attention to the ocean, which has a tremendous number of biological and mineral resources waiting to be developed
A SiamRPN++ underwater target tracking algorithm based on the inverted residual bottleneck block is proposed, which overcomes the problems of poor definition and contrast, complex and changeable background environments, and occluded targets, and reduces the computational complexity of the algorithm
The robustness is reduced from 0.507 to 0.415, the anti-interference ability is increased by 18.15%, the total number of missing reduced from 11 to 9, the average frame rate from 48.02 to 73.74 with an increase of 53.56%, the network complexity from 4.856 to 3.013, the damping is 37.95%, the number of network parameters from 16.220 to 12.538, the damping is 22.70%, and the EAO increased from 0.177 to 0.303 with an increase of 71.19%
Summary
In order to cope with the depletion of land resources and the deterioration of the ecological environment, people have turned their attention to the ocean, which has a tremendous number of biological and mineral resources waiting to be developed. Bazeille et al [6] proposed a color-based light attenuation method which is used to detect and track underwater targets. Z. Wang et al.: Underwater Single Target Tracking Method Using SiamRPN++ Based on Inverted Residual Bottleneck Block path matches the real path, but its speed and accuracy were poor. Current convolutional neural network algorithms, including MDNet [13], SINT [14], and GOTURN [15] These algorithms are used to open-air target tracking, and widely used in many fields such as visual surveillance [16], human-computer interaction [17], and augmented reality [18]. Since the deep separable convolution uses a single-channel filter, the architecture is small in size, less computationally expensive, and high in accuracy, which makes it possible for underwater embedded devices to run neural network models.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
