| 车前小型障碍物图像检测与分类方法 |
| |
| 引用本文： | 陈炳煌.车前小型障碍物图像检测与分类方法[J].福建工程学院学报,2020,0(1):57-62. |
| |
| 作者姓名： | 陈炳煌 |
| |
| 作者单位： | 福建工程学院信息科学与工程学院 |
| |
| 摘 要： | 针对车辆辅助驾驶系统中遇到的障碍物小的特点和对实时性的高要求，提出一种基于卷积神经网络YOLO图像检测算法优化并增加分类计数的方法。通过对小石子和道路坑洞这2种极易引发车辆事故的典型小型障碍物图像建立数据库，针对数据库利用k-Means+优化k值并配置新的锚定值，对取自车载视频的图像进行检测识别。新增的分类和计数算法可快速、直观地获得检测结果，实现驾驶员快速决策的目标。实验结果表明，该方法可对小石子和道路坑洞等小型障碍物有效地检测识别和分类计数，检测速度也满足系统的实时性要求。
|
| 关 键 词： | 目标检测 小型障碍物 YOLO 分类 |
Image detection and classification of small obstacles in front of vehicles |
| |
| CHEN Binghuang.Image detection and classification of small obstacles in front of vehicles[J].Journal of Fujian University of Technology,2020,0(1):57-62. |
| |
| Authors: | CHEN Binghuang |
| |
| Affiliation: | School of Information Science and Engineering, Fujian University of Technology |
| |
| Abstract: | In view of the small obstacles encountered in the vehicle’s assisted driving system and the high real-time requirements, an optimized method based on convolutional neural network YOLO image detection algorithm was proposed and the classification counting was added. By establishing a database for images of such typical small obstacles as small stones and road potholes, which tend to cause vehicle accidents, the images from the vehicle video were detected and identified by using k-means+ to optimize k value and configure new anchors for the database. The newly added classification and counting algorithm can obtain test results quickly and intuitively and realize the goal of rapid driver decision-making. Experimental results show that this method can effectively detect, identify, classify and count small obstacles such as small stones and road potholes, and the detection speed also meets the real-time requirements of the system. |
| |
| Keywords: | object detection small obstacles YOLO classification |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《福建工程学院学报》浏览原始摘要信息 |
|
点击此处可从《福建工程学院学报》下载全文 |
