Abstract
Ship detection is one of the fundamental tasks in computer vision. In recent years, the methods based on convolutional neural networks have made great progress. However, improvement of ship detection in aerial images is limited by large-scale variation, aspect ratio, and dense distribution. In this paper, a Critical and Align Feature Constructing Network (CAFC-Net) which is an end-to-end single-stage rotation detector is proposed to improve ship detection accuracy. The framework is formed by three modules: a Biased Attention Module (BAM), a Feature Alignment Module (FAM), and a Distinctive Detection Module (DDM). Specifically, the BAM extracts biased critical features for classification and regression. With the extracted biased regression features, the FAM generates high-quality anchor boxes. Through a novel Alignment Convolution, convolutional features can be aligned according to anchor boxes. The DDM produces orientation-sensitive feature and reconstructs orientation-invariant features to alleviate inconsistency between classification and localization accuracy. Extensive experiments on two remote sensing datasets HRS2016 and self-built ship datasets show the state-of-the-art performance of our detector.
Highlights
Ship detection is a technology aiming at distinguishing and locating ships of interest
To address the misalignment problem in single-stage detectors and extract critical features for detection, we propose a Critical and Align Feature Constructing Network (CAFC-Net) consisting of three modules: a Bias Attention Module (BAM), a Feature Alignment Module (FAM), and a Distinctive Detection Module (DDM). e Biased Attention Module (BAM) can generate different critical features for classification and regression tasks. e FAM employs an Anchor Optimization Network (AON) and an Alignment Convolution (ACL)
With the BAM used alone, the detection performance is improved by 2.79%, indicating that the critical features constructed by the BAM can facilitate the matching of anchor boxes
Summary
Ship detection is a technology aiming at distinguishing and locating ships of interest. These methods have achieved great success, their performance will decline sharply when they are used for ship detection in aerial images. The red anchor boxes do not capture any critical features necessary for ship detection Despite their high localization accuracy, these anchor boxes reduce the detection performance and should preferably be discarded. Based on the critical features for regression tasks, the AON allows highquality anchor boxes to be generated. With the BAM and the FAM embedded, we design a light single-stage detector which can generate biased critical features for classification and regression and align features for accurate ship detection in aerial images.
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.
