Power Distribution Lines object recognition Using IBM Cloud
Abstract
Visual recognition of objects in an image and the ability to classify them is one of the main areas of research using artificial intelligence. There are currently several frameworks available to assist in this task. IBM Cloud is one such framework in which it offers a broad catalog of tools in the field of artificial intelligence, one of which is IBM Watson Visual Recognition. A model was created through this API using Deep Learning algorithms to analyze images. The main objective of the model is the recognition of eight distinct object types in power distribution lines. All images used in the experiment were collected in approximately 10 different regions of the city. The results obtained in the model created in the article were compared with IBM’s general classifier. The hit rate was at least one correctly recognized component in 33 of the 40 test images, ie correctly classified 82.5% of the test image set, with an accuracy of over 50% for the recognized object. The services used for the whole experiment are from free lite plan. Thus, this article demonstrates the capability of the tool even with the limitations of this version.
References
S. S. Damaceno and R. O. Vasconcelos, “Inteligência artificial: uma breve abordagem sobre seu conceito real e o conhecimento popular,” Revista periódica Caderno de Graduação das Ciências Exatas e Tecnológicas – CGCET, vol. 5, no. 1, pp. 11–16, 2018.
R. L. Stange, “Adaptatividade em aprendizagem de máquina: Conceitos e estudos de caso,” Mestrado em Engenharia Elétrica, Adaptatividade em Aprendizagem de Máquina: Conceitos e Estudos de Caso, São Paulo, 2011.
I. Cloud. Catálogo. [Online]. Available: https://cloud.ibm.com/catalog?category=ai
I. W. Studio. Ibm watson visual recognition. [Online]. Available: https://www.ibm.com/br-pt/cloud/watson-visual-recognition
Y. LeCun and C. Cortes, “MNIST handwritten digit database,” 2010. [Online]. Available: http://yann.lecun.com/exdb/mnist/
