Expansion and Containerization of Pre-trained Deep Learning Models for the Stanford Dogs Data Set

Emil RYBACZEWSKI-JLASSI, Krzysztof DŁUGOKĘCKI, Rafał SZADKOWSKI, Paweł KACZMAREK and Zbigniew PIOTROWSKI

Abstract:

This article presents the characteristic features of the Stanford Dogs data set and discusses the latest deep models of neural networks which were subject to expansion. The method of selection and the use of the best model were presented for the data set in a practical implementation using the containerization of the entire runtime environment of services recognizing dog breeds.