Road Marking Detection Based on Mask R-CNN Instance Segmentation Model
Tian Jin, Jiazheng Yuan, Hongzhe Liu
Abstract
In recent years, deep learning has made a lot of progress in the field of image segmentation. In the intelligent driving, high accuracy in road marking detection is required. Traditional detection methods need manual adjustment parameters and face many difficulties. It is still challenging to design a robust detection algorithm. Based on this, we propose a road marking model detection method. Our method based on the Mask R-CNN for road image instance segmentation to improve precision of detecting road marking. The experimental results show that the method can robustly extract road markers under various complex traffic scene.
Topics & Concepts
Artificial intelligenceComputer scienceSegmentationComputer visionImage segmentationFace (sociological concept)Deep learningField (mathematics)Intelligent transportation systemImage (mathematics)Object detectionPattern recognition (psychology)Robustness (evolution)EngineeringMathematicsSocial scienceCivil engineeringBiochemistryChemistrySociologyPure mathematicsGeneAdvanced Neural Network ApplicationsIndustrial Vision Systems and Defect DetectionVehicle License Plate Recognition