Publication: Analysis of the Possibility of the Neural Network Implementation of the Viola-Jones Algorithm
Дата
2020
Авторы
Journal Title
Journal ISSN
Volume Title
Издатель
Аннотация
© 2020, Springer Nature Switzerland AG.The practice of using the Viola-Jones algorithm and its modifications to solve the problem of finding objects of interest (OI) in the image frame is analyzed. It is shown that the Viola-Jones algorithm is usually used in conjunction with other algorithms in order to solve a complex task-searching for OI, identifying and analyzing its characteristic features. The relevance of the unification of the applied computing means for solving the above complex problem is underlined. The main computational procedures of the Viola-Jones algorithm are considered: obtaining the integral form of the representation of the input image frame, processing the Haar features (HF), implementation of the cascade classifier. Presented and analyzed options for building a neural network (NN) for the implementation of these procedures. The possibility of expanding the functionality of the Viola-Jones algorithm in its implementation based on the NN is shown. The results of an experimental approbation approach are considered in solving the problem of isolating a person’s face with the subsequent isolation of the eye and pupil areas in order to assess their motor activity.
Описание
Ключевые слова
Цитирование
Kolobashkina, L. V. Analysis of the Possibility of the Neural Network Implementation of the Viola-Jones Algorithm / Kolobashkina, L.V., Alyushin, M.V. // Advances in Intelligent Systems and Computing. - 2020. - 948. - P. 232-239. - 10.1007/978-3-030-25719-4_30