Abstract:Due to the large variations of environment with ever-changing background and vehicles with different shapes, colors and appearances, to implement a real-time on-board vehicle recognition system with high adaptability, efficiency and robustness in complicated environments, remains challenging. This paper introduces a simultaneous detection and tracking framework for robust on-board vehicle recognition based on monocular vision technology. The framework utilizes a novel layered machine learning and particle filter to build a multi-vehicle detection and tracking system. In the vehicle detection stage, a layered machine learning method is presented, which combines coarse-search and fine-search to obtain the target using the AdaBoost-based training algorithm. The pavement segmentation method based on characteristic similarity is proposed to estimate the most likely pavement area. Efficiency and accuracy are enhanced by restricting vehicle detection within the downsized area of pavement. In vehicle tracking stage, a multi-objective tracking algorithm based on target state management and particle filter is proposed. The proposed system is evaluated by roadway video captured in a variety of traffics, illumination, and weather conditions. The evaluating results show that, under conditions of proper illumination and clear vehicle appearance, the proposed system achieves 91.2% detection rate and 2.6% false detection rate. Experiments compared to typical algorithms show that, the presented algorithm reduces the false detection rate nearly by half at the cost of decreasing 2.7%–8.6% detection rate. This paper proposes a multi-vehicle detection and tracking system, which is promising for implementation in an on-board vehicle recognition system with high precision, strong robustness and low computational cost.

Multi-vehicle Detection Algorithm Through Combining Harr and HOG Features.

Multi-vehicle detection algorithm through combining Harr and HOG features

A Method for Vehicle Detection Based on Local Gradients Vector

Vehicle detection from complex scenes based on combination features and ELM

Multi-Object Vehicle Detection and Tracking Algorithm Based on Improved YOLOv8 and ByteTrack

Research On Multi-Feature Front Vehicle Detection Algorithm Based On Video Image

Enhanced Detection Model and Joint Scoring Strategy for Multi-Vehicle Tracking

New Method of Moving Vehicle Detection Based on Partial HOG Feature

Detection and Classification of Moving Vehicle from Video Using Multiple Spatio-Temporal Features

The improved spatial histogram features and integrated system for vehicle detection

A Hybrid Vehicle Detection Method Based on Viola-Jones and HOG + SVM from UAV Images

Research on video vehicle detection based on AdaBoost classifiers of the ROI

Vehicle Detection From Aerial Photographing Under Different Illumination and Pose

A Multi-object Detection Method Based on Connected Vehicles

A Vehicle Tracking Algorithm Combining Detector and Tracker

Research and Implementation of Vehicle Tracking Algorithm Based on Multi-Feature Fusion.

An HMM-Based Algorithm for Vehicle Detection in Congested Traffic Situations

Vehicle Detection Algorithm Based on Future Fusion and Improved ObjectNess Binarized Normed Gradients Technology

AdaBoost-Bagging Vehicle Detection Algorithm Based on Multi-Mode Weak Classifier

Simultaneous Multi-Vehicle Detection and Tracking Framework with Pavement Constraints Based on Machine Learning and Particle Filter Algorithm

Real-Time Vehicle Object Detection Method Based on Multi-Scale Feature Fusion