DEVELOPMENT AND RESEARCH OF THE MODEL FOR VIDEO INFORMATION CLASSIFICATION

Abstract

The article is devoted to solving the scientific problem of classifying video content in the face of an increase in the information volume. Computer vision is a very relevant field of artificial intelligence technologies application to expand the capabilities of various search and archive systems. The authors give definitions to the main terms of the studied subject area. A formalized statement of the problem to be solved is presented. A detailed classification of possible options for solving the problem is given. With the rapid development of information technology, digital content is showing an explosive growth trend. The classification of sports videos is of great importance for archiving digital content on the server. Many data mining and machine learning algorithms have made great strides in many application areas (such as classification, regression, and clustering). However, most of these algorithms have a common drawback when the training and test samples are in the same feature space and follow the same distribution. This article discusses the importance of solving the problem of the video information content classification and automatic annotation, and also develops a model based on deep learning and big data. As part of this study, the authors developed a model that improves the quality of video classification, which improves search results. The results of the computational experiment show that the proposed model can be effectively used to classify video events within the sports subject area based on the use of a convolutional neural network. At the same time, high accuracy of sports training video classification is provided. Compared with other models, the proposed model has the advantages of simple implementation, fast processing speed, high classification accuracy, and high generalization ability.