Characters Defect Detection Method of Meter Based on Deep Learning

Aiming at the problems of low accuracy and low efficiency of characters defect detection in traditional meter display screen, a step-by-step detection method based on deep learning was designed. Firstly, the images collected by the camera were marked with features,the block algorithm of characters area in meter was used to detect meters, which realized the accurate classification of the characters area of the meter. Then, the defect features were marked for each region characters after cutting, and the algorithm of characters defect detection was used to detect the nine areas, which solved the problem of low detection rate of meter characters defects. Finally, the proposed method was verified by experiments. The results show that the block accuracy can reach 99.9%, the speed can reach 0.6 s/picture, the characters defect detection accuracy can reach 98%, the speed can reach 1.04 s/block, and the detection time of a meter is 1.64 s. The proposed method can meet the requirements of detection accuracy and time in the actual production of meters.