Electric energy meter electricity utilization information identification algorithm based on computer vision technology

Abstract

According to an electric energy meter electricity utilization information identification algorithm based on the computer vision technology, the framework integrates detection and identification, and end-to-end text positioning and prediction are achieved. Firstly, a detection end performs feature extraction on an input image by combining a feature pyramid network and a residual network, generates a Bezier curve through four control points to better fit a textbox; then, a recognition end adopts a text recognition algorithm based on a convolutional recurrent neural network, introduces a gating recurrent unit to replace a long and short term memory unit, and recognizes a target region text in combination with an attention mechanism; and finally, five groups of ablation experiments are carried out, and performance comparison and evaluation analysis are carried out through experimental data. Experimental results show that the recognition precision of the algorithm is as high as 99.08%, the reasoning speed is high, and the algorithm can be applied to practical application of electricity consumption information detection and recognition.

Description

Technical field [0001] The present invention belongs to the field of dynamic energy genetic information technology using computer vision technologies, and in particular, to the electrical energy meter electrical information recognition algorithm based on adaptive Bessel curve network. Background technique [0002] With the continuous development of China's artificial intelligence technology and the transformation and expansion of the power industry, the electric energy table acts as a metering tool for obtaining electricity information, playing an important role in people's daily lives. The traditional artificial meter reading method is relatively susceptible to the influence of external instability. Therefore, quickly and accurately obtain the electricity information for electricity-related department information. [0003] Typically, the meter is arranged neatly, and the camera lens is perpendicular to the dial plane, and the image of the electric meter is acquired by photograph...

AI Technical Summary

Problems solved by technology

Methods for detection based on manually designed features [1-2] It has high recognition accuracy when the light is sufficient, and when the environment is not ideal, the recognition accuracy will become worse

Therefore, the method based on manually designed features not only consumes a lot of manpower and time, but also takes a long time to train, and the actual detection effect is not ideal.

Images