摘要
货车装载状态高清视频监控系统每日产生的图像数量众多,目前完全依靠人工识别异常状态,工作量巨大且效率低下。采集货车装载状态图像进行人工分类和定义后,输入深度学习模块进行学习,得到正常及异常状态模式识别算法。基于此算法开发的铁路货车装载状态图像智能识别系统可替代大部分人工查看的工作量,明显提高效率,进一步保证安全。
The high-definition video surveillance system for wagon loading conditions produces mass data every day. At present, the identification of abnormality relies on human work, which means huge workload and low efficiency. This paper proposes a normal and abnormal condition identification algorithm by collecting, classifying and defining the wagon loading condition images and inputting the data in the deep learning module. The image intelligent identification system for railway freight wagon loading condition developed based on this algorithm can take over most of the workload previously taken by human and thus improves the efficiency and safety.
作者
张萼辉
ZHANG Ehui(Science and Technology Research Institute, Shanghai Railway Administration, Shanghai 200071, China)
出处
《中国铁路》
2017年第9期113-116,共4页
China Railway
关键词
铁路货车
装载状态
异常检测
图像智能识别
深度学习模块
Railway freight wagon
loading condition
abnormality detection
image intelligent identhfication
deep learning module
