Mapping pigments in paintings with hyperspectral reflectance image cubes using AI
Speaker: Dr John K. Delaney (National Gallery of Art, Washington, US) and Dr Tania Kleynhans (Rochester Institute of Technology, New York, US)
Date: 21/05/2021
Abstract: Spectral imaging modalities, such as reflectance spectroscopy, have been thought to be useful in helping to identify and map pigments in paintings given that many pigments have a unique spectral signature in the visible and infrared. The direct classification and labeling of pigments remain challenging since many paints are intimate pigment mixtures that require a non-linear unmixing model for a robust solution. In this paper we review the science behind reflectance imaging spectroscopy, and results of experiments using a one-dimensional (spectral) convolutional neural network to do direct classification and labeling of pigments in 3-D data cubes of paintings.
chair: Dr. Simone Parisotto
Stay in touch: https://www.turing.ac.uk/research/int...
Subscribe to the youtube channel: AI&ARTS Alan Turing Institute — YouTube