A method for detection and identification of a pointer water meter

Abstract

The invention discloses a method for detecting and identifying a pointer water meter in the technical field of pattern recognition and artificial intelligence, including a data collection terminal and a recognition server. The data collection terminal collects water meter images and uploads the collected images to the recognition server , the identification server performs the detection and identification of the water meter readings, and the processing steps of the identification server include: data preprocessing, label making of training data, construction of deep convolutional neural network, training of neural network model and other steps. The invention uses a camera to collect water meter images, without changing the existing water meter metering equipment, can realize the collection of water meter data, has low cost, good scalability, and is convenient for the connection of metering water meters of different specifications, so it has good practical application value.

Description

technical field [0001] The invention relates to the technical fields of pattern recognition and artificial intelligence, in particular to a detection and recognition method of a pointer water meter based on a deep convolutional neural network. Background technique [0002] With the development of artificial intelligence technology, especially the application of deep network model technology in the field of computer vision, automatic detection and recognition technology based on deep network model has become one of the current hot technologies. In particular, the recent rapid development of GPU and other hardware technologies has solved the computing bottleneck of deep networks to a certain extent, making the practical application of deep neural networks possible. [0003] Water data collection (meter reading) is a tedious and important task for water energy operation units. At present, the method of manual meter reading is mainly adopted: on the one hand, the manual meter r...

AI Technical Summary

Problems solved by technology

At present, the method of manual meter reading is mainly adopted: on the one hand, the manual meter reading mode requires a lot of manpower and material resources; on the other hand, because the meter reading work involves thousands of households, data collection cannot be completed in time. There will also be a series of security risks

Another mode of data collection is digital transformation of existing water meters to realize automatic data collection. The main problems of this mode are: (1) The specifications of meters vary greatly, and the reconstruction of meters involves a large area and requires a lot of capital investment; (2) ) In the water department, mechanical meters have the advantages that electronic equipment such as no need for power supply, safety and reliability cannot be replaced; (3) The cost of new digital meters is relatively high, and maintenance is also difficult

For example, the current photoelectric direct-reading water meter has complicated manufacturing and installation processes, and the cost has increased significantly.

Images