Smart Meter Reader

Over the past few years, many systems that used to operate manually have been replaced by the automated systems. Smart devices provide the opportunity to track dynamic information about different environmental impacts and develop an understanding of human activities and their interactions. Smart tec

Project Title

Smart Meter Reader

Project Area of Specialization

Artificial Intelligence

Project Summary

Over the past few years, many systems that used to operate manually have been replaced by the automated systems. Smart devices provide the opportunity to track dynamic information about different environmental impacts and develop an understanding of human activities and their interactions. Smart technology may collectively enable new applications across a wide variety of domains including image processing, data and information security, authentication and access control. Electricity, water and gas companies follow a very complex procedure for meter reading and bill generation. The current technology uses manual way of the reading meter in which an operator has to note reading of meter and also take a picture of the meter as a proof. In order to generate the bill, operator has to confirm the reading; the operator has to perform a tedious work of matching the reading with the snap. This bill generating technique is very lengthy and is open to errors. Our aim is to propose a methodology to create an efficient system that would automate the existing system and will reduce the time consumed and human effort. Reading data efficiently from the meter is our major goal. We exploit optical character algorithm and region-based techniques e.g. Region based convolutional neural network to read data from the images and used firebase to securely transfer data to the server. We have developed an android application that is connected to the firebase and it uses an optical character recognition algorithm to read data from the images and securely sent it to the server. We have evaluated our designed system at each step as well as on the overall system. The problem is complex but our proposed methodologies tend to produce an efficient result when they are applied to real-time images. Our propose methodologies provide an efficient solution to the problems related to meter reading.

Project Objectives

The aim of our project is to provide a methodology to automate the existing meter reading system. Following are the objectives that we want to achieve:  

• To propose a methodology for reading data from meter using an image processing technique.
• To securely send the collected data to the server.
• To implement an android application for the proposed methodology.

Project Implementation Method

We are using Faster R-CNN model for the extraction of the Roi. We are using separate models for localization and recognition. The second model will be used for the recognition of the numeric characters. We hope to deliver two solutions. We are going to use GPU to train the model. For the second solution, we need a raspberry pi, a neural compute stick and a camera.

Benefits of the Project

The main benefits of the project are:

• It will save the company a lot of time.
• It will save the company the cost of hiring new employees.
• It will greatly reduce the chances of mistake in the reading.

Technical Details of Final Deliverable

Smart meter reader's Android app has been developed using Android studio and Firebase database. We have used optical character recognition algorithm for reading numbers from meters and then we will send the extracted data to the firebase using a secure connection where it will be saved along with the other information of every user. We also developed a handheld device that consists of raspberry pi and a camera that can also be used to detect numbers from the meter. We have established an algorithm using convolutional neural networks and burned that algorithm on a raspberry pi that detects and extracts digits from the meter with great accuracy this alternative is better than optical character recognition one as that one is fast but not very much accurate. To conclude we have estabished two solution for our problem one is an android application for recoginition of readings from meter and sending it to server and a handheld device that uses deep learning algorithms for detecion and recoginition of meter readings.

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

IT

Technologies

Artificial Intelligence(AI)

Sustainable Development Goals

Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com