Application of random forest algorithm to classify vehicles detected by a multiple inductive loop system

This paper presents a suitable algorithm to classify vehicles detected by a multiple inductive loop system, developed for measuring traffic parameters in a heterogeneous and no-lane disciplined traffic. The proposed classification scheme employs Random Forest (RF) algorithm. This scheme is suited not only for classifying the detected vehicles as bicycle, motorcycle, scooter, car and bus but also for counting them accurately under a mixed traffic condition. The algorithm has been implemented and tested. Its performance has also been compared with other algorithms based on threshold values and signature patterns. The threshold, signature and RF based algorithms use the number of loops a vehicle occupies as an important factor for classification. Results from a prototype system developed show that the RF based algorithm provides better accuracy compared to the threshold based and signature based methods.

[1]  M. Pursula,et al.  Microprocessor and PC-based vehicle classification equipments using induction loops , 1989 .

[2]  S. S. M. Ali,et al.  A magnetically coupled inductive loop sensing system for less-lane disciplined traffic , 2012, 2012 IEEE International Instrumentation and Measurement Technology Conference Proceedings.

[3]  R. Sroka,et al.  VEHICLE CLASSIFICATION BY PARAMETRIC IDENTIFICATION OF THE MEASURED SIGNALS , 2000 .

[4]  Carlos Sun.  An Investigation in the Use of Inductive Loop Signatures for Vehicle Classification , 2000 .

[5]  Boby George,et al.  A simple multiple loop sensor configuration for vehicle detection in an undisciplined traffic , 2011, 2011 Fifth International Conference on Sensing Technology.

[6]  Doo-Kwon Baik,et al.  Vehicle-Classification Algorithm for Single-Loop Detectors Using Neural Networks , 2006, IEEE Transactions on Vehicular Technology.

[7]  Muhammed Gökhan Cinsdikici,et al.  Vehicle-Classification Algorithm Based on Component Analysis for Single-Loop Inductive Detector , 2010, IEEE Transactions on Vehicular Technology.

[8]  Antanas Verikas,et al.  Mining data with random forests: A survey and results of new tests , 2011, Pattern Recognit..

[9]  A multiple loop vehicle detection system for heterogeneous and lane-less traffic , 2011, 2011 IEEE International Instrumentation and Measurement Technology Conference.

[10]  R. Sroka.  Data fusion methods based on fuzzy measures in vehicle classification process , 2004, Proceedings of the 21st IEEE Instrumentation and Measurement Technology Conference (IEEE Cat. No.04CH37510).

[11]  Heng Wei,et al.  Artificial neural network method for length-based vehicle classification using single-loop outputs , 2006 .

[12]  R. L. Anderson.  Electromagnetic loop vehicle detectors , 1970 .