Abstract
This paper presents a Random Repeatable Network (RRN) which is an entirely unsupervised training framework for interest point detection. The existing learning based methods make tradeoff between the unsupervised level and the approximation degree to the objective of repeatability, while our RRN model trains a convolutional neural network whose loss function is directly based on point repeatability without relying on any initial interest point detector. In terms of point repeatability under perspective transform or illumination change, we propose a novel loss function with regularization term of repeatability, which is optimized by an effective iterative algorithm. Experiments demonstrate our model achieves better performance on test data compared to some state-of-the-art interest point detector.
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.
