Research on the Application of Text Box Recognition Algorithm Based on Convolutional Neural Network in Power Service System

A text box recognition algorithm based on a convolutional neural network (CNN) is proposed to address the low efficiency in information input of business terminals in the power industry. A Faster RCNN network is used to train and validate the text box dataset, and combined with OCR technology to develop an auxiliary input system. By introducing a CNN-based text box recognition algorithm, the algorithm’s applicability is improved for business terminal applications across different systems without changing the original system architecture. Experimental results show that the CNN-based text box recognition algorithm applied to the auxiliary input system significantly improves information input speed and accuracy compared to manual input methods and has broad application prospects in business terminals in the power industry.