Abstract
Automatic ship detection, recognition, and counting are crucial for intelligent maritime surveillance, timely ocean rescue, and computer-aided decision-making. YOLOv3 pretraining model is used for model training with sample images for ship detection. The ship detection model is built by adjusting and optimizing parameters. Combining the target HSV color histogram features and LBP local features’ target, object recognition and selection are realized by using the deep learning model due to its efficiency in extracting object characteristics. Since tracking targets are subject to drift and jitter, a self-correction network that composites both direction judgment based on regression and target counting method with variable time windows is designed, which better realizes automatic detection, tracking, and self-correction of moving object numbers in water. The method in this paper shows stability and robustness, applicable to the automatic analysis of waterway videos and statistics extraction.
Highlights
Target tracking has great demand in computer vision research in recent years
Traditional target tracking algorithms generally use the target’s color, texture, contour, gradient histogram, Haar, SIFT, SURF, and other single features to represent the target in the process of target appearance modeling [1]
In order to solve these problems mentioned above, this paper proposes a selfcorrecting ship target tracking and counting method with variable time window based on YOLOv3
Summary
Target tracking has great demand in computer vision research in recent years. Traditional target tracking algorithms generally use the target’s color, texture, contour, gradient histogram, Haar, SIFT, SURF, and other single features to represent the target in the process of target appearance modeling [1]. Luca Bertinetto et al [13] proposed SimFC, a target tracking algorithm based on deep learning, to track targets through a fully convolutional twin network. Yang and Chan [17] applied the residual network to target tracking and proposed the CREST algorithm to perform residual learning by detecting the difference between the extracted convolution features and the real data of the target object. YOLOv3 Target Detection Algorithm e YOLOv3 target detection algorithm is an end-to-end target detection algorithm, which is of fast speed and high accuracy, which meets the requirements of real-time detection [20] It uses a fully convolutional neural network based on the Darknet-53 network to extract image features. All candidate targets come from the targets detected by this YOLOv3 target detection algorithm in the follow-up tracking
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: 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.