Research Article
D2TFRS: An Object Recognition Method for Autonomous Vehicles Based on RGB and Spatial Values of Pixels
@INPROCEEDINGS{10.1007/978-3-319-94180-6_16, author={Furqan Alam and Rashid Mehmood and Iyad Katib}, title={D2TFRS: An Object Recognition Method for Autonomous Vehicles Based on RGB and Spatial Values of Pixels}, proceedings={Smart Societies, Infrastructure, Technologies and Applications. First International Conference, SCITA 2017, Jeddah, Saudi Arabia, November 27--29, 2017, Proceedings}, proceedings_a={SCITA}, year={2018}, month={7}, keywords={Autonomous driving Autonomous vehicles Object recognition Decision tree Decision fusion Deep learning Majority voting C5.0 SVM}, doi={10.1007/978-3-319-94180-6_16} }- Furqan Alam
Rashid Mehmood
Iyad Katib
Year: 2018
D2TFRS: An Object Recognition Method for Autonomous Vehicles Based on RGB and Spatial Values of Pixels
SCITA
Springer
DOI: 10.1007/978-3-319-94180-6_16
Abstract
Autonomous driving is now near future reality which will transform our world due to its numerous benefits. The foremost challenge to this task is to correctly identify the objects in the driving environment. In this work, we propose an object recognition method known as Decision Tree and Decision Fusion based Recognition System (D2TFRS) for autonomous driving. We combined two separate feature sets, which are RGB pixel values and spatial points X,Y of each pixel to form our dataset. The D2TFRS is based on our intuition that reclassification of pre-identified misclassified objects in a driving environment can give better prediction accuracy. Results showed that D2TFRS outperformed AdaBoost classifier and performed better than C5.0 classifier in terms of the classification accuracy and Kappa. In terms of speed, C5.0 outperforms both AdaBoost and D2TFRS. However, D2TFRS outperformed AdaBoost with respect to speed. We strongly believe that D2TFRS will have better parallelization performance compared to the other two methods and it will be investigated in our future work.
