A Detection Algorithm Design based on Image Segmentation Strategy for Vehicle Damage Recognition
- Resource Type
- Conference
- Authors
- Min Htet Thar; Cheol Min Lee; Dong Myung Lee
- Source
- 한국통신학회 학술대회논문집. 2021-11 2021(11):462-463
- Subject
- Language
- Korean
- ISSN
- 2383-8302
In this paper, a detection algorithm design based on image segmentation strategy for vehicle damage recognition was proposed by constructing a vehicle segmentation dataset to analyze damage recognition for each body part of a vehicle. The proposed algorithm is processed based on the Mask RCNN (mask region of convolutional network) model and includes convolutional neural network (CNN), region proposal network (RPN) and region of interest (RoI). The proposed algorithm is expected to be able to act on behalf of humans in the field of self-driving vehicle damage estimation and auto insurance claim process after sufficient testing and reliability process.