PL EN


Preferencje help
Widoczny [Schowaj] Abstrakt
Liczba wyników
Tytuł artykułu

Weryfikacja autentyczności kolorów na zdjęciach wykonanych w technice analogowej

Treść / Zawartość
Identyfikatory
Warianty tytułu
EN
Checking color authenticity in analog photos
Konferencja
Zastosowanie komputerów w nauce i technice 2019 (XXIX ; 2019 ; Gdańsk ; Polska)
Języki publikacji
PL
Abstrakty
PL
W artykule opisano zagadnienie odróżniania historycznych fotografii pomiędzy oryginalnie kolorowe a koloryzowane. Rozważono problem doboru zdjęć pod względem technologii, w jakiej zostały wykonane. Następnie wykorzystując sieci neuronowe już w części wyuczone na innych zbiorach danych, sprawdzono ich efektywność w rozwiązywaniu badanego problemu. Rozważono wpływ rozmiaru obrazu podanego na wejściu, architektury zastosowanej sieci, a także zestawu danych użytego do uczenia sieci i wyodrębniania cech. W rezultacie potwierdzono przydatność opracowanego zbioru do treningu sieci, a także zaobserwowano, że zwiększanie rozmiaru sieci nie przynosi dodatkowych korzyści. Uzyskana trafność rozróżniania sięgnęła ponad 92 %.
EN
The article describes a dataset designed to train neural networks distinguishing historical photographs between the ones that have original historic color and those which were contemporary colorized. The problem of choosing photos in terms of technology and content was considered. Using some of the pre-trained neural networks on other collections, their effectiveness in solving the studied issue was checked. The influence of the input image size, the depth of the neural network used as well as the data set used to train the network to extract features was investigated. As a result, the usefulness of the developed set for network training was confirmed, and it was observed that increasing the network did not bring any additional benefits. The reached accuracy is up to 92.6%.
Twórcy
  • Politechnika Gdańska, Wydział Elektroniki, Telekomunikacji i Informatyki Katedra Systemów Multimedialnych
Bibliografia
  • 1. O. Russakovsky i in., „ImageNet Large Scale Visual Recognition Challenge”, Int. J. Comput. Vis., t. 115, nr 3, ss. 211-252, 2015.
  • 2. G. Van Horn i in., „The iNaturalist Species Classification and Detection Dataset – Supplementary Material”, Comput. Vis. Found., nr July, ss. 4-6, 2017.
  • 3. S. Maji, E. Rahtu, J. Kannala, M. Blaschko, i A. Vedaldi, „Fine-Grained Visual Classification of Aircraft”, 2013.
  • 4. T. Gebru, J. Krause, Y. Wang, D. Chen, J. Deng, i F. F. Li, „Fine-grained car detection for visual census estimation”, 31st AAAI Conf. Artif. Intell. AAAI 2017, ss. 4502–4508, 2017.
  • 5. J. Krause, M. Stark, J. Deng, i L. Fei-Fei, „3D object representations for fine-grained categorization”, w Proceedings of the IEEE International Conference on Computer Vision, 2013, ss. 554–561.
  • 6. L. Yang, P. Luo, C. C. Loy, i X. Tang, „A large-scale car dataset for fine-grained categorization and verification”, w Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2015, t. 07-12-June, ss. 3973–3981.
  • 7. M. Domański, Obraz cyfrowy: reprezentacja, kompresja, podstawy przetwarzania: standardy JPEG i MPEG. Wydawnictwa Komunikacji i Łączności, 2010.
  • 8. P. Tångeberg, „The use of colours on the seventeenth century royal warship Vasa”, Endeavour, t. 24, nr 4. ss. 147–151, 2000.
  • 9. C.-A. Francois-Franck, „Colour microphotography with autochrome plates by MM. A. et L. Lumiere.”, COMPTES RENDUS Hebd. DES SEANCES L Acad. DES Sci., t. 144, ss. 1340–1342, 1907.
  • 10. L. D. Mannes i L. Godowsky, „The Kodachrome Process for Amateur Cinematography in Natural Colors”, J. Soc. Motion Pict. Eng., 1935.
  • 11. M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, i L. C. Chen, „MobileNetV2: Inverted Residuals and Linear Bottlenecks”, w Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2018, ss. 4510–4520.
  • 12. K. He, X. Zhang, S. Ren, i J. Sun, „Identity mappings in deep residual networks”, w Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2016, t. 9908 LNCS, ss. 630-645.
  • 13. C. Szegedy, V. Vanhoucke, S. Ioffe, i J. Shlens, „Rethinking the Inception Architecture for Computer Vision”.
  • 14. J. Donahue i in., „DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition”.
Uwagi
Opracowanie rekordu ze środków MNiSW, umowa Nr 461252 w ramach programu "Społeczna odpowiedzialność nauki" - moduł: Popularyzacja nauki i promocja sportu (2020).
Typ dokumentu
Bibliografia
Identyfikator YADDA
bwmeta1.element.baztech-cc6a6828-eca8-48f8-af6f-61e64eb537fa
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.