Johan van Zyl

Maria Spichkova, Johan van Zyl

In this report we analysed a possibility of using computer vision techniques for automated reading of utility meters. In our study, we focused on two computer vision techniques: an open-source solution Tensorflow Object Detection (Tensorflow) and a commercial solution Anyline. This report extends our previous publication: We start with presentation of a structured analysis of related approaches. After that we provide a detailed comparison of two computer vision technologies, Tensorflow Object Detection (Tensorflow) and Anyline, applied to semi-automated reading of utility meters. In this paper, we discuss limitations and benefits of each solution applied to utility meters reading, especially focusing on aspects such as accuracy and inference time. Our goal was to determine the solution that is the most suitable for this particular application area, where there are several specific challenges.

Maria Spichkova, Johan van Zyl, Siddharth Sachdev et al.

Electricity and gas meter reading is a time consuming task, which is done manually in most cases. There are some approaches proposing use of smart meters that report their readings automatically. However, this solution is expensive and requires (1) replacement of the existing meters, even when they are functional and new, and (2) large changes of the whole system dealing with the meter readings. This paper presents results of a project on automation of the meter reading process for the standard (non-smart) meters using computer vision techniques, focusing on the comparison of two computer vision techniques, Google Cloud Vision and AWS Rekognition.