Convolutional Neural Networks Used to Date Photographs
Nowadays, the information provided by digital photographs is very complete and very relevant in different professional fields, such as scientific or forensic photography. Taking this into account, it is possible to determine the date when they were taken, as well as the type of device that they were...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-01-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/11/2/227 |
| _version_ | 1797494546512216064 |
|---|---|
| author | Jesús-Ángel Román-Gallego María-Luisa Pérez-Delgado Sergio Vicente San Gregorio |
| author_facet | Jesús-Ángel Román-Gallego María-Luisa Pérez-Delgado Sergio Vicente San Gregorio |
| author_sort | Jesús-Ángel Román-Gallego |
| collection | DOAJ |
| description | Nowadays, the information provided by digital photographs is very complete and very relevant in different professional fields, such as scientific or forensic photography. Taking this into account, it is possible to determine the date when they were taken, as well as the type of device that they were taken with, and thus be able to locate the photograph in a specific context. This is not the case with analog photographs, which lack any information regarding the date they were taken. Extracting this information is a complicated task, so classifying each photograph according to the date it was taken is a laborious task for a human expert. Artificial intelligence techniques make it possible to determine the characteristics and classify the images automatically. Within the field of artificial intelligence, convolutional neural networks are one of the most widely used methods today. This article describes the application of convolutional neural networks to automatically classify photographs according to the year they were taken. To do this, only the photograph is used, without any additional information. The proposed method divides each photograph into several segments that are presented to the network so that it can estimate a year for each segment. Once all the segments of a photograph have been processed, a general year for the photograph is calculated from the values generated by the network for each of its segments. In this study, images taken between 1960 and 1999 were analyzed and classified using different architectures of a convolutional neural network. The computational results obtained indicate that 44% of the images were classified with an error of less than 5 years, 20.25% with a marginal error between 5 and 10 years, and 35.75% with a higher marginal error of more than 10 years. Due to the complexity of the problem, the results obtained are considered good since 64.25% of the photographs were classified with an error of less than 10 years. Another important result of the study carried out is that it was found that the color is a very important characteristic when classifying photographs by date. The results obtained show that the approach given in this study is an important starting point for this type of task and that it allows placing a photograph in a specific temporal context, thus facilitating the work of experts dedicated to scientific and forensic photography. |
| first_indexed | 2024-03-10T01:35:52Z |
| format | Article |
| id | doaj.art-d12f81970c554371bd79f324cb540650 |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-10T01:35:52Z |
| publishDate | 2022-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-d12f81970c554371bd79f324cb5406502023-11-23T13:34:19ZengMDPI AGElectronics2079-92922022-01-0111222710.3390/electronics11020227Convolutional Neural Networks Used to Date PhotographsJesús-Ángel Román-Gallego0María-Luisa Pérez-Delgado1Sergio Vicente San Gregorio2Higher Polytechnic School of Zamora, University of Salamanca, Avenida de Requejo 33, 49022 Zamora, SpainHigher Polytechnic School of Zamora, University of Salamanca, Avenida de Requejo 33, 49022 Zamora, SpainHigher Polytechnic School of Zamora, University of Salamanca, Avenida de Requejo 33, 49022 Zamora, SpainNowadays, the information provided by digital photographs is very complete and very relevant in different professional fields, such as scientific or forensic photography. Taking this into account, it is possible to determine the date when they were taken, as well as the type of device that they were taken with, and thus be able to locate the photograph in a specific context. This is not the case with analog photographs, which lack any information regarding the date they were taken. Extracting this information is a complicated task, so classifying each photograph according to the date it was taken is a laborious task for a human expert. Artificial intelligence techniques make it possible to determine the characteristics and classify the images automatically. Within the field of artificial intelligence, convolutional neural networks are one of the most widely used methods today. This article describes the application of convolutional neural networks to automatically classify photographs according to the year they were taken. To do this, only the photograph is used, without any additional information. The proposed method divides each photograph into several segments that are presented to the network so that it can estimate a year for each segment. Once all the segments of a photograph have been processed, a general year for the photograph is calculated from the values generated by the network for each of its segments. In this study, images taken between 1960 and 1999 were analyzed and classified using different architectures of a convolutional neural network. The computational results obtained indicate that 44% of the images were classified with an error of less than 5 years, 20.25% with a marginal error between 5 and 10 years, and 35.75% with a higher marginal error of more than 10 years. Due to the complexity of the problem, the results obtained are considered good since 64.25% of the photographs were classified with an error of less than 10 years. Another important result of the study carried out is that it was found that the color is a very important characteristic when classifying photographs by date. The results obtained show that the approach given in this study is an important starting point for this type of task and that it allows placing a photograph in a specific temporal context, thus facilitating the work of experts dedicated to scientific and forensic photography.https://www.mdpi.com/2079-9292/11/2/227photographartificial intelligenceconvolutional neural networksimage recognitionclassification |
| spellingShingle | Jesús-Ángel Román-Gallego María-Luisa Pérez-Delgado Sergio Vicente San Gregorio Convolutional Neural Networks Used to Date Photographs Electronics photograph artificial intelligence convolutional neural networks image recognition classification |
| title | Convolutional Neural Networks Used to Date Photographs |
| title_full | Convolutional Neural Networks Used to Date Photographs |
| title_fullStr | Convolutional Neural Networks Used to Date Photographs |
| title_full_unstemmed | Convolutional Neural Networks Used to Date Photographs |
| title_short | Convolutional Neural Networks Used to Date Photographs |
| title_sort | convolutional neural networks used to date photographs |
| topic | photograph artificial intelligence convolutional neural networks image recognition classification |
| url | https://www.mdpi.com/2079-9292/11/2/227 |
| work_keys_str_mv | AT jesusangelromangallego convolutionalneuralnetworksusedtodatephotographs AT marialuisaperezdelgado convolutionalneuralnetworksusedtodatephotographs AT sergiovicentesangregorio convolutionalneuralnetworksusedtodatephotographs |