Abstract:
Most of the previous works had used SVM or AdaBoost to select the best features and thresholds for object classification. This paper uses statistical and geometrical feat...Show MoreMetadata
Abstract:
Most of the previous works had used SVM or AdaBoost to select the best features and thresholds for object classification. This paper uses statistical and geometrical features for LiDAR-based vehicle detection. We train our system using labeled point sets, which we obtain by manually labeled on KITTI dataset. We also define 11 features of LIDAR point set as input features. According to the test, the correct detection rate is above 98.3%. Experimental results show that our approach can perform accurately and fast, work well in most of cases under our experimental environments. The more data we use, the more accurate experimental result we can obtain.
Date of Conference: 13-15 December 2016
Date Added to IEEE Xplore: 09 February 2017
ISBN Information:
Electronic ISSN: 2474-2325