Convolutional Neural Networks for Automatic Meter Reading


1. Paper Information

1.1. Authors

Rayson Laroca, Victor Barroso, Matheus A. Diniz, Gabriel R. Gonçalves, William Robson Schwartz, David Menotti.

1.2. Abstract

In this paper, we tackle Automatic Meter Reading (AMR) by leveraging the high capability of Convolutional Neural Networks (CNNs). We design a two-stage approach that employs the Fast-YOLO object detector for counter detection and evaluates three different CNN-based approaches for counter recognition. In the AMR literature, most datasets are not available to the research community since the images belong to a service company. In this sense, we introduce a new public dataset, called UFPR-AMR dataset, with 2,000 fully and manually annotated images. This dataset is, to the best of our knowledge, three times larger than the largest public dataset found in the literature and contains a well-defined evaluation protocol to assist the development and evaluation of AMR methods. Furthermore, we propose the use of a data augmentation technique to generate a balanced training set with many more examples to train the CNN models for counter recognition. In the proposed dataset, impressive results were obtained and a detailed speed/accuracy trade-off evaluation of each model was performed. In a public dataset, state-of-the-art results were achieved using less than 200 images for training.

1.3. Citation

If you use the UFPR-AMR dataset or the annotations provided by us in your research, please cite our paper:

  • R. Laroca, V. Barroso, M. A. Diniz, G. R. Gonçalves, W. R. Schwartz, D. Menotti, “Convolutional Neural Networks for Automatic Meter Reading,” Journal of Electronic Imaging, vol. 28, no. 1, p. 013023, 2019. [SPIE Digital Library] [PDF] [BibTeX] [Copyright Notice]

You may also be interested in our new research, where we proposed an end-to-end approach for AMR in unconstrained scenarios and introduced the Copel-AMR dataset:

  • R. Laroca, A. B. Araujo, L. A. Zanlorensi, E. C. de Almeida, D. Menotti, “Towards Image-based Automatic Meter Reading in Unconstrained Scenarios: A Robust and Efficient Approach,” IEEE Access, vol. 9, pp. 67569-67584, 2021. [Webpage] [IEEE Xplore] [PDF] [BibTeX]


2.1. UFPR-AMR Dataset

We introduced a public dataset that includes 2,000 images (with 10,000 manually labeled digits) from electric meters of different types and in different conditions. It is three times larger than the largest dataset found in the literature for this task and contains a well-defined evaluation protocol, allowing a fair comparison of different methods.

Full details regarding the dataset, including download instructions, can be seen here.

2.1. Annotations of Public Datasets

We also labeled the region containing the significant digits in two public datasets that have no annotations or contain labels only for part of the pipeline. These annotations are provided along with the UFPR-AMR dataset.

3. Related publications

A list of all papers on AMR published by us can be seen here.

4. Contact

Please contact the first author ( with questions or comments.