A Mobile Recognition System for Analog Energy Meter Scanning
In December, our CTO Daniel and our Computer Vision Expert Martin got the chance to fly to Las Vegas for the 12th International Symposium on Visual Computing! There they talked about the paper they wrote together with the researcher Gayane Shalunts from SAIL LABS on “A mobile Recognition System for Analog Energy Meter Scanning”. This paper got published in the Lecture Notes in Computer Science Series by Springer.
Get in touch with Anyline and find out how to improve your processes today!
The paper presents a mobile platform based system, scanning electricity, gas and water meters. Optical Character Recognition is therefore used to automate manual procedures. Electricity, gas and water meters are being scanned with the smartphone instead of typing the meter value in manually for example. This process of using OCR speeds up the procedure and automates it while increasing the accuracy. Furthermore the human effort and error is decreased. Occurring problems were for example bad lighting conditions, different font types on the meter or variable digit size and spacing.
The approached method consists of the two main stages of digit detection and OCR. The Optical Character Recognition Technology is built by using Google Tesseract and Convolutional Neural Networks. Those two approaches are compared to each other.
Throughout the studies, Convolutional Neural Network significantly outperform the Tesseract OCR for all types of meters.
Here you can find the whole paper online!
QUESTIONS? LET US KNOW!
Sign up for our Newsletter!
Be the first to receive the latest news, release notes and updates!