基于深度学习的海上升压站仪表识别方法

Deep Learning-based Instrument Recognition Technology for Sea Ascending Pressure Station

随着人工智能和自动化水平的提高,机器人被越来越广泛地应用于变电站、升压站等场所巡检,为图像采集带来了便利,但是仪表图像的自动读数识别准确性问题尚未很好解决。提出一种基于深度学习的仪表识别技术,将仪表识别过程简化为仪表检测和仪表识别两个子任务,针对子任务的特点和难度设计不同的深度学习模型,通过某海上升压站现场采集的仪表图片对模型进行训练和验证,结果表明提出的模型具有较高的准确度和识别速度。

With the improvement of artificial intelligence and automation level,robots are more and more widely used in substations,booster stations and other places to inspect,which brings convenience to image collection,but the accuracy of automatic reading and recognition of instrument images is not very good.This paper proposes an instrument recognition technology based on deep learning,which simplifies the process of instrument identification into two sub-tasks of instrument detection and instrument identification.Different deep learning models are designed according to the characteristics and difficulty of the sub-tasks.The instrument pictures are used to train and verify the model,and the results show that the model proposed in this paper has high accuracy and recognition speed.