A coarse-to-fine approach for vehicles detection from aerial images

Vehicles detection in aerial images has a wide range of applications for visual surveillance. This paper introduces a framework for robust on-road vehicle detection. A passively trained framework system is built using conventional supervised learning. The strategy which is proposed for detecting vehicles is From-coarse-to-fine. In the first step. Road is segmented with LSD algorithm to narrow the area which will be detected. AdaBoost based algorithm is used for coarse detection. SVM is used to reduce false rates. Experimental results show that this framework yields a efficient and robust on-board vehicle detection system with high precision and low false rates.