An Improved YOLOv7-Based Model for Real-Time Meter Reading with PConv and Attention Mechanisms

Xiancheng Peng,Yangzhuo Chen,Xiaowen Cai,Jun Liu

DOI: https://doi.org/10.3390/s24113549

IF: 3.9

2024-05-31

Sensors

Abstract:With the increasing complexity of the grid meter dial, precise feature extraction is becoming more and more difficult. Many automatic recognition solutions have been proposed for grid meter readings. However, traditional inspection methods cannot guarantee detection accuracy in complex environments. So, deep-learning methods are combined with grid meter recognition. Existing recognition systems that utilize segmentation models exhibit very high computation. It is challenging to ensure high real-time performance in edge computing devices. Therefore, an improved meter recognition model based on YOLOv7 is proposed in this paper. Partial convolution (PConv) is introduced into YOLOv7 to create a lighter network. Different PConv introduction locations on the base module have been used in order to find the optimal approach for reducing the parameters and floating point of operations (FLOPs). Meanwhile, the dynamic head (DyHead) module is utilized to enhance the attention mechanism for the YOLOv7 model. It can improve the detection accuracy of striped objects. As a result, this paper achieves mAP50val of 97.87% and mAP50:90val of 62.4% with only 5.37 M parameters. The improved model's inference speed can reach 108 frames per second (FPS). It enables detection accuracy that can reach ±0.1 degrees in the grid meter.

engineering, electrical & electronic,instruments & instrumentation,chemistry, analytical

What problem does this paper attempt to address?

This paper aims to solve the problem of real-time and accurate recognition of electricity grid meter readings. The accuracy of traditional methods in complex environments cannot be guaranteed. Therefore, this paper proposes an improved model based on YOLOv7, which combines Partial Convolution (PConv) and DyHead Attention mechanism. By introducing PConv into the basic module of YOLOv7 to reduce network parameters and floating-point operations (FLOPs), and utilizing DyHead to enhance the detection accuracy of bar-shaped objects. The experimental results show that the improved model achieves 97.87% mAPval50 and 62.4% mAPval50:90 while maintaining 5.37M parameters. It can also achieve an inference speed of 108 frames per second and accurately detect the angle of the electricity meter to ±0.1 degrees. The paper also proposes an adaptive calibration algorithm to enhance the robustness of pointer reading calculations. This model is suitable for meter reading in complex environments such as lighting changes and reflections, especially in handling blurry, dim light, and small target scenarios, it demonstrates stable recognition capabilities.