Отрывок: These methods were evaluated on the real task of emotion recognition from frontal and 45° rotated facial images detected in the widely used Radboud Faces Database (RaFD) [26]. The neural networks were trained from scratch using identical training samples and learning procedures. Inspired by the well-known facial expression recognition CNN [27], we choose VGG-S architecture for the HashedNet, BWN and XNOR-Net as a baseline. All the neural network model...
Полная запись метаданных
Поле DC | Значение | Язык |
---|---|---|
dc.contributor.author | Rassadin, A.G. | - |
dc.contributor.author | Savchenko, A.V. | - |
dc.date.accessioned | 2017-05-15 12:44:12 | - |
dc.date.available | 2017-05-15 12:44:12 | - |
dc.date.issued | 2017 | - |
dc.identifier | Dspace\SGAU\20170515\63757 | ru |
dc.identifier.citation | Rassadin A.G. Deep neural networks performance optimization in image recognition / A.G. Rassadin, A.V. Savchenko // Сборник трудов III международной конференции и молодежной школы «Информационные технологии и нанотехнологии» (ИТНТ-2017) - Самара: Новая техника, 2017. - С. 649-654. | ru |
dc.identifier.uri | http://repo.ssau.ru/handle/Informacionnye-tehnologii-i-nanotehnologii/Deep-neural-networks-performance-optimization-in-image-recognition-63757 | - |
dc.description.abstract | In this paper, we consider the problem of insufficient runtime and memory-space complexities of contemporary deep convolutional neural networks in the problem of image recognition. A survey of recent compression methods and efficient neural networks architectures is provided. The experimental study is focused on the visual emotion recognition problem. We compare the computational speed and memory consumption during the training and the inference stages of such methods as the weights matrix decomposition, binarization and hashing in the visual emotion recognition problem. It is experimentally shown that the most efficient recognition is achieved with the full network binarization and matrices decomposition. | ru |
dc.description.sponsorship | The article was prepared within the framework of the Academic Fund Program at the National Research University Higher School of Economics (HSE) in 2017 (grant №17-05-0007) and by the Russian Academic Excellence Project "5-100". A.V. Savchenko is supported by Russian Federation President grant no. МД-306.2017.9. | ru |
dc.language.iso | en | ru |
dc.publisher | Новая техника | ru |
dc.subject | deep neural networks | ru |
dc.subject | image recognition | ru |
dc.subject | visual emotion recognition | ru |
dc.subject | deep compression | ru |
dc.subject | binarized neural networks | ru |
dc.subject | tensor train | ru |
dc.subject | tensor decomposition | ru |
dc.subject | XNOR-Net | ru |
dc.title | Deep neural networks performance optimization in image recognition | ru |
dc.type | Article | ru |
dc.textpart | These methods were evaluated on the real task of emotion recognition from frontal and 45° rotated facial images detected in the widely used Radboud Faces Database (RaFD) [26]. The neural networks were trained from scratch using identical training samples and learning procedures. Inspired by the well-known facial expression recognition CNN [27], we choose VGG-S architecture for the HashedNet, BWN and XNOR-Net as a baseline. All the neural network model... | - |
Располагается в коллекциях: | Информационные технологии и нанотехнологии |
Файлы этого ресурса:
Файл | Описание | Размер | Формат | |
---|---|---|---|---|
paper 122_649-654.pdf | Основная статья | 560.46 kB | Adobe PDF | Просмотреть/Открыть |
Показать базовое описание ресурса
Просмотр статистики
Поделиться:
Все ресурсы в архиве электронных ресурсов защищены авторским правом, все права сохранены.