Browsing by Author "Chen, Ailin"
Now showing 1 - 4 of 4
Results Per Page
Sort Options
- Authentication of Art: assessing the performance of a machine learning based authentication methodPublication . Chen, Ailin; Jesus, Rui; Vilarigues, MárciaThis paper compares the test results generated by applying the method for the authentication of paintings by Portuguese artist Amadeo de Souza Car-doso in the interest of exploring the generalisation properties of the algorithm on other artists or genres. This sets the base for the method to be improved and de-veloped accordingly in future applications for a broader audience in a wider set-ting. The obtained results show that the classifier obtained from the algorithm using paintings appears not to be directly applicable to drawings of the same art-ist. When the classifier is retrained for a different genre like Chinese paintings or artists like van Gogh, the algorithm appears to perform as well as the classifier on Amadeo paintings, i.e. the algorithm is sufficient for the classification of a specific type of artist or genre.
- Convolutional neural network-based pure paint pigment identification using hyperspectral imagesPublication . Chen, Ailin; Jesus, Rui; VilariguesThis research presents the results of the implementation of deep learning neural networks in the identification of pure pigments of heritage artwork, namely paintings. Our paper applies an inno vative three-branch deep learning model to maximise the correct identification of pure pigments. The model proposed combines the feature maps obtained from hyperspectral images through multi ple convolutional neural networks, and numerical, hyperspectral metric data with respect to a set of reference reflectances. The results obtained exhibit an accurate representation of the pure pre dicted pigments which are confirmed through the use of analytical techniques. The model presented outperformed the compared coun terparts and is deemed to be an important direction, not only in terms of utilisation of hyperspectral data and concrete pigment data in heritage analysis, but also in the application of deep learning in other fields.
- Hyperspectral image reconstruction of heritage artwork using RGB images and deep neural networksPublication . Chen, Ailin; Jesus, Rui; Vilarigues, M.The application of our research is in the art world where the scarcity of available analytical data from a particular artist or physical access for its acquisition is restricted. This poses a fundamental problem for the purpose of conservation, restoration or authentication of historical artworks. We address part of this problem by providing a practical method to generate hyperspectral data from readily available RGB imagery of artwork by means of a two-step process using deep neural networks. The particularities of our approach include the generation of learnable colour mixtures and reflectances from a reduced collection of prior data for the mapping and reconstruction of hyperspectral features on new images. Further analysis and correction of the prediction are achieved by a second network that reduces the error by producing results akin to those obtained by a hyperspectral camera. Our method has been used to study a collection of paintings by Amadeo de Souza-Cardoso where successful results were obtained. CCS CONCEPTS • Computing methodologies → Neural networks; Artificial intelligence; • Applied computing → Arts and humanities.
- Using deep learning techniques for authentication of Amadeo de Souza Cardoso paintings and drawingsPublication . Chen, Ailin; Jesus, Rui; Villarigues, MárciaThis paper investigates the application of a Convolutional Neural Network (CNN), AlexNet, on the authentication of paintings by different artists, including Portuguese painter Amadeo de Souza Cardoso, Chinese painter Daqian Zhang and Dutch painter Vincent van Gogh. The research is motivated by the studies on the identification of the works by Amadeo based on the painter’s brushstroke implementing Machine Learning algorithms combined with material analysis. The employment of CNN intends to improve the performance of the brushstroke analysis and increase the accuracy while authenticating an artist’ works. The results show that the implementation of AlexNet produces higher accuracies than its counterparts applying previous brushstroke analysis. Notably, when Amadeo drawings are included in the testing based on Amadeo paintings, the accuracies obtained with the original algorithm drop substantially, whilst the counterparts attained with AlexNet improved considerably. However, when other testing sets are introduced, especially the Chinese paintings, the accuracies show a great increase with the original algorithm but a significant decrease with AlexNet. It implies that AlexNet surpasses the traditional computation through learning by examples; it can potentially solve the problem of limited number of artworks by a specific artist for training.