樊玉平 孙子正 李雅芹 赵海月
（南京南自信息技术有限公司，江苏 南京 210000）
摘要：为进一步保障新能源发电场站作业人员安全，设计一套融合双模态目标检测的作业人员安全管控系统。该系统采用融合预处理后图像和目标分割图像的双模态目标检测方法对作业人员及其附属物进行目标检测，并利用分割一切模型提高目标分割精度。实验结果表明，该系统提升了作业人员及其附属物的检测准确率，提高了作业人员行为分析识别的可靠性，可满足电力行业作业人员安全管控的高标准需求。
关键词：融合双模态；目标检测；安全管控；分割一切模型；新能源发电场站
中图分类号：TP391 文献标志码：A 文章编号：1674-2605(2026)02-0007-07
DOI：10.12475/aie.20260207 开放获取
Design of a Worker Safety Management and Control System Integrating Dual-modality Object Detection
FAN Yuping SUN Zizheng LI Yaqin ZHAO Haiyue
(Nanjing Nanzi Information Technology Co., Ltd., Nanjing 210000, China)
Abstract: To further ensure the safety of workers in new energy power plants, a worker safety control system integrating dual-modality object detection is designed. The system employs a dual-modality object detection method that combines preprocessed images and segmented object images to detect workers and their accessories, and utilizes the segment anything model to improve the accuracy of object segmentation. Experimental results show that the system improves the detection accuracy of workers and their accessories, enhances the reliability of worker behavior analysis and recognition, and meets the high standards of worker safety control in the electric power industry.
Keywords: dual-modality fusion; object detection; safety management and control; segment anything model; new energy power plants
