As enhance of safety requirement in the smart cities, biometrics recognition, as an approach for society safety, has been greatly researched and developed. The identification of working status of welders will help judge whether they are wearing personal protective equipment correctly. We proposed an improved algorithm based on SSD (Single Shot Multibox Detector) that can identify three mainstream manual welding methods including SMAW (shielded metal arc welding), GMAW (gas metal arc welding) and TIG (tungsten inert gas), which has never been researched before and can promote the intelligentization of welding monitoring to construct smart cities. The improvement includes two parts. Firstly, the backbone of SSD is replaced with MobileNetV3. Then, a feature fusion module is added to enhance the information of low-level feature maps to improve detection accuracy. The experimental results of our welding behavior detector show that, the mAP is 87.45%, detection speed is 25FPS, and parameter memory is 87.5 MB, which has a relatively excellent performance considering speed, accuracy and memory.

随着智慧城市对安全性要求的提高，生物识别技术作为一种社会安全手段已经得到了很大的研究和发展。识别焊工的工作状态将有助于判断他们是否正确穿戴了个人防护设备。我们提出了一种基于SSD（单发多盒检测器）的改进算法，该算法可以识别三种主流的手工焊接方法，包括SMAW（保护金属电弧焊），GMAW（气体金属电弧焊）和TIG（钨极惰性气体），这是从未有过的。之前的研究，可以促进焊接监控的智能化，建设智慧城市。改进包括两个部分。首先，SSD的骨干被MobileNetV3取代。然后，增加了特征融合模块，增强了低级特征图的信息，提高了检测精度。我们的焊接行为检测器的实验结果表明，mAP为87.45％，检测速度为25FPS，参数存储为87.5 MB，在速度，精度和存储方面都具有相对优异的性能。