An Improved YOLOv2 for Vehicle Detection.

Jun Sang,Zhongyuan Wu,Qian Zhang,Hong Xiang,Haibo Hu,Pei Guo,Bin Cai

doi:10.3390/s18124272

Abstract

Vehicle detection is one of the important applications of object detection in intelligent transportation systems. It aims to extract specific vehicle-type information from pictures or videos containing vehicles. To solve the problems of existing vehicle detection, such as the lack of vehicle-type recognition, low detection accuracy, and slow speed, a new vehicle detection model YOLOv2_Vehicle based on YOLOv2 is proposed in this paper. The k-means++ clustering algorithm was used to cluster the vehicle bounding boxes on the training dataset, and six anchor boxes with different sizes were selected. Considering that the different scales of the vehicles may influence the vehicle detection model, normalization was applied to improve the loss calculation method for length and width of bounding boxes. To improve the feature extraction ability of the network, the multi-layer feature fusion strategy was adopted, and the repeated convolution layers in high layers were removed. The experimental results on the Beijing Institute of Technology (BIT)-Vehicle validation dataset demonstrated that the mean Average Precision (mAP) could reach 94.78%. The proposed model also showed excellent generalization ability on the CompCars test dataset, where the “vehicle face” is quite different from the training dataset. With the comparison experiments, it was proven that the proposed method is effective for vehicle detection. In addition, with network visualization, the proposed model showed excellent feature extraction ability.

Highlights

In order to properly solve urban traffic problems and overcome the existing disadvantages, such as the lack of enough vehicle information and the low accuracy of vehicle information retrieval, intelligent transportation was strongly developed
References [1,2,3] adopted the histogram of oriented gradient (HOG) method to extract vehicle-type features in images, and classified those features using the support vector machine (SVM), achieving vehicle detection
Considering that the down-sampling factor was 32, all randomly selected input image sizes were multiples of 32, where the minimum size was 352 × 352 and the maximum size was 608 × 608. Such a training method enables the final model to better predict the images with different sizes, while the same model can be used for vehicle detection with different resolutions, which may enhance the robustness of the model

Summary

Introduction

In order to properly solve urban traffic problems and overcome the existing disadvantages, such as the lack of enough vehicle information and the low accuracy of vehicle information retrieval, intelligent transportation was strongly developed. As an indispensable part of this method, vehicle detection is widely studied by researchers all over the world. The common vehicle detection methods can be divided into two categories: traditional methods and deep-learning-based methods. The traditional methods refer to traditional machine learning algorithms. References [1,2,3] adopted the histogram of oriented gradient (HOG) method to extract vehicle-type features in images, and classified those features using the support vector machine (SVM), achieving vehicle detection. In Reference [4], a deformable part model (DPM)

Methods

Results

Conclusion