01627563

Component

https://doi.org/10.3141/2645-13

In recent years there has been growing interest in the use of nonintrusive systems such as radar and infrared systems for vehicle recognition. State-of-the-art nonintrusive systems can report up to eight classes of vehicle types. Video-based systems, which arguably are the most popular nonintrusive detection systems, can report only very coarse classification levels (up to four classes), even with the best-performing vision systems. The present study developed a vision system that can report finer vehicle classifications according to FHWA’s scheme and is also comparable to other nonintrusive recognition systems. The proposed system decoupled object recognition into two main tasks: localization and classification. It began with localization by generating class-independent region proposals for each video frame, then it used deep convolutional neural networks to extract feature descriptors for each proposed region, and, finally, the system scored and classified the proposed regions by using a linear support vector machines template on the feature descriptors. The precision of the system varied by vehicle class. Passenger cars and SUVs were detected at a precision rate of 95%. The precision rates for single-unit, single-trailer, and double-trailer trucks ranged between 92% and 94%. According to receiver operating characteristic curves, the best system performance can be achieved under free flow, daytime or nighttime, and with good video resolution.

01628860

17-02006

English

pp 113–122

2017

Transportation Research Record: Journal of the Transportation Research Board

9780309460408

Digital/other

Figures (9) ; References (23)

Automatic vehicle detection and identification systems; Intelligent vehicles; Neural networks

Object recognition

Data and Information Technology; Highways; Operations and Traffic Management

PRP, TRIS, TRB, ATRI

Dec 8 2016 10:42AM

• Application of Dempster–Shafer Data Fusion Technique in Support of Decision Making with Big Data
• Data Processing Framework for Development of Driving Cycles with Data from SHRP 2 Naturalistic Driving Study
• Data-Based Sorting Algorithm for Variable Message Sign Location: Case Study of Beijing
• Detecting Imminent Lane Change Maneuvers in Connected Vehicle Environments
• Detection of Crack Growth in Asphalt Pavement Through Use of Infrared Imaging
• Development and Application of an Ecodriving Support Platform Based on Internet+: Case Study in Beijing Taxicabs
• Economic Feasibility Study for Pavement Monitoring Using Synthetic Aperture Radar-Based Satellite Remote Sensing: Cost–Benefit Analysis
• Enhancing Relative Positioning Accuracy in Vehicle-to-Vehicle Safety: Multivehicle Kinematics and Nonlinear Particle Filtering Approach
• Evaluating the Promotion of Innovative Mobility Options with an Agent-Based Modeling and Simulation Approach
• Hidden Markov Approach to Dynamically Modeling Car Ownership Behavior
• Hunt for Perfect Detection of Parking Occupancy: Evaluation of On-Street Technology and Its Ability to Address Urban Challenges
• Investigating Impacts of Communication Loss on Signal Performance with Use of Event-Based Data
• Methodology for Calculating Latency of GPS Probe Data
• Multicontextual Machine-Learning Approach to Modeling Traffic Impact of Urban Highway Work Zones
• Novel Cooperative Collision Avoidance Model for Connected Vehicles
• Optimized Video Tracking for Automated Vehicle Turning Movement Counts
• Short-Term Traffic Flow Forecasting for Urban Roads Using Data-Driven Feature Selection Strategy and Bias-Corrected Random Forests
• Traffic Congestion Pattern Classification Using Multiclass Active Shape Models
• Trajectory Segmentation Map-Matching Approach for Large-Scale, High-Resolution GPS Data
• Vehicle Trajectory Analysis System via Mutual Information and Sparse Reconstruction