ArtInsight: A detailed dataset for detecting deterioration in easel paintingsZenodo
Francisco M. Garcia-Moreno,
Jose Manuel del Castillo de la Fuente,
Luis Rodrigo Rodríguez-Simón,
María Visitación Hurtado-Torres
Affiliations
Francisco M. Garcia-Moreno
Department of Software Engineering, Computer Science School, University of Granada, C/Periodista Daniel Saucedo Aranda, 18014, Spain; Research Centre for Information and Communication Technologies (CITIC-UGR), University of Granada, 18014, Spain; Corresponding author at: Department of Software Engineering, Computer Science School, University of Granada, C/Periodista Daniel Saucedo Aranda, 18014, Spain.
Jose Manuel del Castillo de la Fuente
Department of Paint and Restoration, Faculty of Fine Arts, University of Granada, Av. Andalucía n° 38, 18071, Spain
Luis Rodrigo Rodríguez-Simón
Department of Paint and Restoration, Faculty of Fine Arts, University of Granada, Av. Andalucía n° 38, 18071, Spain
María Visitación Hurtado-Torres
Department of Software Engineering, Computer Science School, University of Granada, C/Periodista Daniel Saucedo Aranda, 18014, Spain; Research Centre for Information and Communication Technologies (CITIC-UGR), University of Granada, 18014, Spain
ArtInsight is an innovative dataset designed to detect deterioration in fine art, specifically easel paintings. The dataset includes high-resolution images captured at the University of Granada using a digital camera with a 105 mm lens, ISO 125, F5, and a shutter speed of 1/13, and processed for color calibration. Two types of images are featured: those showing stucco technique interventions and those with Lacune from the loss of the Painting Layer (LPL). The VGG Image Annotator was employed for manual damage labeling, with annotations exported in JSON format and labeled for stucco and LPL damages. The dataset comprises 14 images with 2909 distinct damage areas, split into training and validation datasets. Developed using Python 3.7 and fine-tuned on a pre-trained Mask-RCNN model, this dataset demonstrates high accuracy rates (98–100 %) in damage detection. ArtInsight aims to facilitate automated damage detection and foster future research in art conservation and restoration. The dataset is publicly available at 10.5281/zenodo.8429814.
