Paper
9 September 2022 Study on target damage assessment method based on mask R-CNN
Mingyan Zhao, Yanan Di, Weilhua Han
Author Affiliations +
Proceedings Volume 12328, Second International Conference on Optics and Image Processing (ICOIP 2022); 123281C (2022) https://doi.org/10.1117/12.2644319
Event: Second International Conference on Optics and Image Processing (ICOIP 2022), 2022, Taian, China
Abstract
Aiming at the difficulty of multi-target damage assessment in complex environment, a target damage assessment method based on Mask R-CNN is proposed. In the method, the pre-training weights are obtained by Mask R-CNN model, and the parameters of the Mask R-CNN backbone network are initialized according to the target size and damage scene. The initialized parameter of Mask R-CNN backbone network is improved based on the pre-training weight. The method solves the problem that a single target can’t provide accurate damage assessment effect. Through simulation experiments of multi-target damage, the reliability and accuracy of the method are improved obviously.
© (2022) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mingyan Zhao, Yanan Di, and Weilhua Han "Study on target damage assessment method based on mask R-CNN", Proc. SPIE 12328, Second International Conference on Optics and Image Processing (ICOIP 2022), 123281C (9 September 2022); https://doi.org/10.1117/12.2644319
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Target detection

Data modeling

Image segmentation

Statistical modeling

Neural networks

Detection and tracking algorithms

Feature extraction

Back to Top