Отрывок: We trained our networks on different number of input detections from 4 to 64. As can be seen in Fig. 5, RMSE errors gradually decrease with input detection number in- creasing, and after detection number reaches 16, this de- crease becomes more gradual. (a) (b) (c) (d) Fig. 5. RMSE for various input detections number: (a) focal length RMSE; (b) tilt RMSE; (c) roll RMSE; (d) height RMSE 5. Evaluation and results We trained every network for 300 ...
Полная запись метаданных
Поле DC Значение Язык
dc.contributor.authorShalimova, E.A.-
dc.contributor.authorShalnov, E.V.-
dc.contributor.authorKonushin, A.S.-
dc.date.accessioned2020-07-30 10:34:35-
dc.date.available2020-07-30 10:34:35-
dc.date.issued2020-06-
dc.identifierDspace\SGAU\20200728\84747ru
dc.identifier.citationShalimova EA, Shalnov EV, Konushin AS. Camera parameters estimation from pose detections. Computer Optics 2020; 44(3): 385-392. DOI: 10.18287/2412-6179-CO-600.ru
dc.identifier.urihttps://dx.doi.org/10.18287/2412-6179-CO-600-
dc.identifier.urihttp://repo.ssau.ru/handle/Zhurnal-Komputernaya-optika/Camera-parameters-estimation-from-pose-detections-84747-
dc.description.abstractSome computer vision tasks become easier with known camera calibration. We propose a method for camera focal length, location and orientation estimation by observing human poses in the scene. Weak requirements to the observed scene make the method applicable to a wide range of scenarios. Our evaluation shows that even being trained only on synthetic dataset, the proposed method outperforms known solution. Our experiments show that using only human poses as the input also allows the proposed method to calibrate dynamic visual sensors.ru
dc.language.isoen_USru
dc.publisherСамарский национальный исследовательский университетru
dc.relation.ispartofseries44/3;-
dc.subjectcamera calibrationru
dc.subjectdynamic vision sensorru
dc.subjectvideo surveillanceru
dc.titleCamera parameters estimation from pose detectionsru
dc.typeArticleru
dc.textpartWe trained our networks on different number of input detections from 4 to 64. As can be seen in Fig. 5, RMSE errors gradually decrease with input detection number in- creasing, and after detection number reaches 16, this de- crease becomes more gradual. (a) (b) (c) (d) Fig. 5. RMSE for various input detections number: (a) focal length RMSE; (b) tilt RMSE; (c) roll RMSE; (d) height RMSE 5. Evaluation and results We trained every network for 300 ...-
Располагается в коллекциях: Журнал "Компьютерная оптика"

Файлы этого ресурса:
Файл Описание Размер Формат  
440309.pdfОсновная статья3.41 MBAdobe PDFПросмотреть/Открыть
Показать базовое описание ресурса Просмотр статистики
Поделиться:



Все ресурсы в архиве электронных ресурсов защищены авторским правом, все права сохранены.