Abstract
Few-Shot Image Segmentation Based on Dual Comparison Module and Sequential k-Shot Integration
Highlights
Image segmentation is one of the basic tasks in the computer vision
Compared with the common segmentation approaches [3,4,5,6], few-shot segmentation focuses on mining the correlation between the query images and the support images, which greatly improves the generalization ability given the limited annotations
The fine comparison submodule of dual comparison module is based on the nonlocal operations and is capable of measuring the dense spatial similarity between the pair of features for the query images and the support images
Summary
Image segmentation is one of the basic tasks in the computer vision. Deep neural networks have significantly promoted its development in recent years. In other image domains such as X-ray images which has the variety of object scales, viewpoints and heavy occlusions, few-shot segmentation may have a performance gap. To overcome these problems, we propose the dual comparison network named DCNet for the task of few-shot segmentation. Experiments illustrate that our model achieves significant performance on few-shot segmentation and has good generalization ability to different data domains. The fine comparison submodule of dual comparison module is based on the nonlocal operations and is capable of measuring the dense spatial similarity between the pair of features for the query images and the support images. The experiment results on our X-ray image dataset demonstrate the generalization on different image domain
Sign-in/Register to view highlights & summary 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 callMore From: International Journal of Computational Intelligence Systems
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.
