基于注意力机制的运动车辆检测方法

Vehicle Attribute Recognition Based on Visual Attention Mechanism

由于视觉注意力机制模仿了人类视觉系统的视觉原理,因此引入计算机视觉注意力机制方法来测试车辆信息是一项被广泛关注的任务。在基于视频广播确定交通信息时,目标标识是一般交通信息的基础。而从交通现场视频中提取目标对于后续的跟踪、分类和数据分析等任务具有十分有效的帮助,从而可以降低算法的复杂性并提高其效率。该文主要着眼于识别道路上的车辆,基于视觉注意力模型,提出了一种基于注意力机制的车辆兴趣区提取系统。通过实验,将该算法与背景差分法的结果进行了比较。

Because the visual attention mechanism simulates the vision principle of human vision system,it is a far-reaching at⁃tempt to introduce the computer vision attention mechanism into the information detection of traffic vehicles.In the traffic informa⁃tion detection based on video stream,the target detection is the basis of the whole traffic information detection.Extracting interest⁃ed objects from traffic scene video is very helpful for the following target tracking,classification and analysis,which can reduce the complexity of calculation and improve efficiency.This paper focuses on the detection of moving vehicles on the road.Based on the visual attention model,this paper proposes a vehicle region of interest extraction method based on the visual attention mechanism.Through experiments,this algorithm is compared and analyzed with the results obtained by the background difference method.