Automatic water meter detection method based on Faster-rcnn

Abstract

The invention provides an automatic water meter detection method based on Faster-rcnn. The method particularly comprises the following steps of 1 data acquiring, wherein 50,000 water meter reading photos are shot with a mobile phone, and multiple water meters (2) are included; 2 data processing, wherein the photos are tailored, and the size of the photos is 480*200 after processing; 3 label making, wherein the water meters are detected by adopting a supervision method, and water meter reading frames are manually calibrated with software; 4 network training, wherein prepared training data and labels are input to a Faster-rcnn network for training; and 5 network testing, wherein testing data is input to the trained network, and finally water meter detection results and probabilities are obtained. According to the method, automatic detection on the water meters is achieved by means of the Faster-rcnn, the water meter reading interval can be quickly detected, the detection time is 0.177 second, the detection accuracy rate reaches 99%, and the extremely high practicability and application value are achieved.

Description

Technical field [0001] The invention belongs to the technical field of pattern recognition and artificial intelligence, and particularly relates to an automatic detection method related to a deep neural network. Background technique [0002] With the rapid development of computer technology, artificial intelligence technology is gradually changing our lives, making our lives more convenient and efficient. The recent rapid development of hardware technologies such as GPUs has also made practical applications of deep neural networks possible. [0003] In real life, we cannot do without tap water, whether in rural or urban areas, tap water has become popular. In the process of tap water settlement, special personnel are required to go to each house every month to copy the water meter readings, which is a time-consuming and laborious task. There are a large number of residential houses, covering a wide range, and even remote, and manual reading of water meter readings is prone to err...

AI Technical Summary

Problems solved by technology

However, during the settlement process of tap water, special personnel are required to go to each household every month to take the readings of the water meter, which is a time-consuming and laborious task.

There are a large number of residential buildings, a wide coverage area, and even remoteness. However, manual reading of water meters is prone to errors. Various factors make manual reading of water meters extremely cumbersome. It is imperative to find an automatic and efficient method of reading water meters.

Images