Machine Learning based Meter Reading Automation

Machine Learning based Meter Reading AutomationFully automated, scalable cloud based solution for fast and accurate odometer reading for vehicle fleets

About Client
Tata Projects is one of the fastest growing and most admired industrial infrastructure companies in India. The organization is part of highly respected Tata Group. Tata projects has expertise in executing large and complex urban and industrial infrastructure projects. They provide ready-to-deploy solutions for refineries, roads, bridges, integrated rail & metro systems, commercial building & airports, and power generation, transmission & distribution systems, chemical process plants, water & waste management, and mining & metal purification systems

Challenge
Tata Projects has several staff that is allocated vehicles or machines that need to be monitored or used for their day-to-day tasks. These machines include such as Heavy Material Handling Equipment, Motor Vehicles, Generators etc. For utilization reconciliation and/or computation of allowances, Tata Projects have provided a mobile application where the staff should enter a start and end odometer reading and submit on the mobile app along with a picture of the evidence of this reading.

The challenges the team faced was:
• Tata Projects has a central team that is responsible to reaudit the updated reading and match whether the Staff entry created has the relevant readings in the images. This task is redundant and monotonous
• The validation process is time consuming – each record may take up to 2-3 minutes which involves downloading images, verifying, and updating the audit tool with the verdicts
• Image uploads are done through a third-party mobile app – Quality of the images are not always clear and readable format
• The Odometer images have several formats for different type of vehicles or machines

Requirement: Create a Machine Learning based solution that achieves the following
• Improves efficiency of the Validation of the Readings received from Staff
• Integrate with the existing mobile platform and submit results
• Publish a report by End Of day for non-compliances i.e., not matching submissions, mismatch of date, expired submissions

Solution
• Highly Scalable Serverless Architecture: Integrate the image storage on AWS S3 for the uploaded images which trigger a Lambda function that invokes various AWS ML services such as Textract, Rekognition
• Interface with the mobile platform using APIs from the platform
• Automated report generation for Accuracy and analysis for the management team
• 96% accuracy of the ML based services which has improved efficiency to 90%

Benefits
• High efficiency with output generated within minutes
• Highly scalable and secure with Serverless infrastructure
• Lowest TCO due to deep expertise & Rapid execution
• Focus on Core Business while we do Heavy Lifting
• Utilize Human Resource for more productive workloads


AWS Services used

cloudmantra used the following AWS Services towards successful project delivery:
• Amazon Sagemaker notebooks to train models and run the algorithm
• AWS Lambda to execute the compute the algorithm for invoking different ML based services for analyzing the images.
• Amazon S3 used to store the Input images uploaded by the representatives. The images are encrypted at rest.
• Amazon Rekogniton is used for analyzing the images for odometer readings
• Amazon Textract is used to extract the Date and Timestamp from Odometer images

Get this case study in your email

Let us deliver the right solution for your business

Your Email (required)