Класифікація об’єктів на зображеннях с використанням бібліотеки tensorflow

Abstract

The given work is devoted to object classification on the images using the neural network. The proposed application can tag objects on the uploaded images by TensorFlow framework model, which was trained on the Coco Dataset. The dataset includes thousands of images with different objects which werelabeled by people. Nodes of the model represent an image processing functions, while the graph edges represent the multidimensional data arrays (tensors) communicated between them. As a result, we have got tagged objects with the percent of the probability of the correct tagging.