Отрывок: Let y = y1, y2,…, yn denote the vector of network outputs of length n from an alphabet A extended with a blank symbol “–”. In training, an objective function that maximizes the log likelihood of target labeling p(w | y) is used. In every training step, the probability p(wgt | y) of every timestamp region in the mini-batch is efficiently calculated using a forward-backward algorithm similar to HMMs t...
Полная запись метаданных
Поле DC | Значение | Язык |
---|---|---|
dc.contributor.author | Cheng, J. | - |
dc.contributor.author | Dai, W. | - |
dc.date.accessioned | 2019-04-08 14:24:21 | - |
dc.date.available | 2019-04-08 14:24:21 | - |
dc.date.issued | 2019-02 | - |
dc.identifier | Dspace\SGAU\20190324\74814 | ru |
dc.identifier.citation | Cheng, J. A framework of reading timestamps for surveillance video / J. Cheng , W. Dai // Computer Optics. - 2019. - Vol. 43, Issue1. - P. 72-77. - DOI: 10.18287/2412-6179-2019-43-1-72-77. | ru |
dc.identifier.uri | https://dx.doi.org/10.18287/2412-6179-2019-43-1-72-77 | - |
dc.identifier.uri | http://repo.ssau.ru/handle/Zhurnal-Komputernaya-optika/A-framework-of-reading-timestamps-for-surveillance-video-74814 | - |
dc.description.abstract | This paper presents a framework to automatically read timestamps for surveillance video. Reading timestamps from surveillance video is difficult due to the challenges such as color variety, font diversity, noise, and low resolution. The proposed algorithm overcomes these challenges by using the deep learning framework. The framework has included: training of both timestamp localization and recognition in a single end-to-end pass, the structure of the recognition CNN and the geometry of its input layer that preserves the aspect of the timestamps and adapts its resolution to the data. The proposed method achieves state-of-the-art accuracy in the end-to-end timestamps recognition on our datasets, whilst being an order of magnitude faster than competing methods. The framework can be improved the market competitiveness of panoramic video surveillance products. | ru |
dc.language.iso | en | ru |
dc.publisher | Самарский национальный исследовательский университет им. акакдемика С.П. Королева, Институт систем обработки изображений РАН - филиал ФНИЦ «Кристаллография и фотоника» РАН | ru |
dc.relation.ispartofseries | 43;1 | - |
dc.subject | surveillance video | ru |
dc.subject | timestamp localization | ru |
dc.subject | timestamp recognition | ru |
dc.title | A framework of reading timestamps for surveillance video | ru |
dc.type | Article | ru |
dc.textpart | Let y = y1, y2,…, yn denote the vector of network outputs of length n from an alphabet A extended with a blank symbol “–”. In training, an objective function that maximizes the log likelihood of target labeling p(w | y) is used. In every training step, the probability p(wgt | y) of every timestamp region in the mini-batch is efficiently calculated using a forward-backward algorithm similar to HMMs t... | - |
Располагается в коллекциях: | Журнал "Компьютерная оптика" |
Файлы этого ресурса:
Файл | Описание | Размер | Формат | |
---|---|---|---|---|
430108.pdf | Основная статья | 624.06 kB | Adobe PDF | Просмотреть/Открыть |
Показать базовое описание ресурса
Просмотр статистики
Поделиться:
Все ресурсы в архиве электронных ресурсов защищены авторским правом, все права сохранены.