Deep Learning for Image-based Automatic Dial Meter Reading: Dataset and Baselines

6 May 2020  ·  Gabriel Salomon, Rayson Laroca, David Menotti ·

Smart meters enable remote and automatic electricity, water and gas consumption reading and are being widely deployed in developed countries. Nonetheless, there is still a huge number of non-smart meters in operation. Image-based Automatic Meter Reading (AMR) focuses on dealing with this type of meter readings. We estimate that the Energy Company of Paran\'a (Copel), in Brazil, performs more than 850,000 readings of dial meters per month. Those meters are the focus of this work. Our main contributions are: (i) a public real-world dial meter dataset (shared upon request) called UFPR-ADMR; (ii) a deep learning-based recognition baseline on the proposed dataset; and (iii) a detailed error analysis of the main issues present in AMR for dial meters. To the best of our knowledge, this is the first work to introduce deep learning approaches to multi-dial meter reading, and perform experiments on unconstrained images. We achieved a 100.0% F1-score on the dial detection stage with both Faster R-CNN and YOLO, while the recognition rates reached 93.6% for dials and 75.25% for meters using Faster R-CNN (ResNext-101).

PDF Abstract

Datasets


Introduced in the Paper:

UFPR-ADMR-v1

Used in the Paper:

ssd

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Meter Reading UFPR-ADMR-v1 YOLOv3 (608 x 608) Rank-1 Recognition Rate 74.75 # 2
Meter Reading UFPR-ADMR-v1 YOLOv3 (416 x 416) Rank-1 Recognition Rate 73.75 # 3
Meter Reading UFPR-ADMR-v1 YOLOv2 (608 x 608) Rank-1 Recognition Rate 71.25 # 6
Meter Reading UFPR-ADMR-v1 YOLOv2 (416 x 416) Rank-1 Recognition Rate 68 # 7
Meter Reading UFPR-ADMR-v1 Fast-YOLOv3 (608 x 608) Rank-1 Recognition Rate 54.25 # 8
Meter Reading UFPR-ADMR-v1 Fast-YOLOv3 (416 x 416) Rank-1 Recognition Rate 47.75 # 10
Meter Reading UFPR-ADMR-v1 Fast-YOLOv2 (608 x 608) Rank-1 Recognition Rate 51.75 # 9
Meter Reading UFPR-ADMR-v1 Fast-YOLOv2 (416 x 416) Rank-1 Recognition Rate 42.25 # 11
Meter Reading UFPR-ADMR-v1 Faster-RCNN (ResNet-101) Rank-1 Recognition Rate 71.75 # 5
Meter Reading UFPR-ADMR-v1 Faster-RCNN (ResNet-50) Rank-1 Recognition Rate 72.25 # 4
Meter Reading UFPR-ADMR-v1 Faster-RCNN (ResNeXt-101) Rank-1 Recognition Rate 75.25 # 1

Methods