Automatic consumption reading on electromechanical meters using HoG and SVM
Darlan B. P. Quintanilha, Robherson W. S. Costa, João Otávio Bandeira Diniz, João Dallyson Sousa de Almeida, Geraldo Bráz, A.C. Silva, Anselmo Cardoso de Paiva, Eliana Márcia Garros Monteiro, Bruno Rodrigues Froz, L.P.A. Piheiro, W. Melho · 2017
The reading of the electrical energy consumption is carried out manually for the most part. One of the ways to validate the measurement performed by the meter reader on the field is through the request of images acquisition in situations where there is a divergence in the consumption informed by the meter reader. However, this validation process in two Brazilian energy companies (CEMAR and CELPA) entails a daily volume of roughly 30 thousand images to be analyzed. In this context, this article presents a method for automatic consumption reading on electromechanical meters based on meter image using HoG and SVM. The method obtained an accuracy of 79.52% in tests performed with more than 8000 digits.