基于CNN的安全智能监测识别算法

Intelligent security monitoring and recognition algorithms based on CNN

针对施工环境的复杂性,监管人员对施工人员着装的监督通常存在着一定的困难,较难实现及时、有效的监督等问题,文中提出了一种基于CNN的安全智能监测识别算法。该算法首先通过相关样本图像训练出所需要的安全帽、安全带等四种识别模型。然后利用所得到的模型,对电力施工现场所拍摄的实时图像进行检测识别,从而实现智能化监测。测试结果表明,该算法对于施工人员着装的平均识别准确率可达到89.27%,验证了该算法的可行性。

Due to the complexity of the construction environment, supervisors often have some difficulties in supervising the construction personnel’s clothing,and it is difficult to achieve real-time and effective supervision. This paper proposes a safety intelligent monitoring and recognition algorithm based on CNN. Firstly,four recognition models,such as safety helmet and safety belt,are trained through the relevant sample images. Then,the real-time images taken at the electric power construction site are detected and recognized by the model,so as to realize intelligent monitoring. The test results show that the average recognition accuracy of the algorithm for construction personnel dress can reach89.27 %,which verifies the feasibility of the algorithm.