Research on Two-step Pointer Meter Recognition Method Based on Yolov7

In order to adapt to the dangerous and complex environment of chemical sites and reduce the danger of manual inspection in high-risk areas, this paper proposes a two-step pointer meter recognition method based on yolov7, which can be combined with explosion-proof inspection robots to complete the autonomous detection of factory meters and meter reading work. By pre-producing the corresponding templates of various types of meters, the yolov7 model is used to complete the identification of the type of meters in the input image, dial area cropping and center area positioning; the cropped dial area image is tilted and corrected, and then processed by filtering and enhancement methods, the improved Hough transform detection method is used to fit a straight line to the location of the meter pointer; the angle of deflection of the pointer is computed and matched to the corresponding template information to complete the identification of the meter number. The angle of pointer deflection is calculated, and the corresponding template information is matched to complete the reading of the meter number. The experimental results show that the method in this paper can achieve good results in both experimental scenes and chemical scenes, and can quickly and accurately recognize a variety of known chemical industry instruments, which can meet the needs of instrument detection in the automatic inspection in chemical plants.