Visual link retrieval and knowledge discovery in painting datasets
This work addresses the need for art historians to analyze visual arts more efficiently, especially in cases with scarce metadata, but it is incremental as it applies existing deep learning and graph analysis methods to a specific domain.
The paper tackles the problem of finding similarity relationships among paintings to aid art historians by presenting a framework for visual link retrieval and knowledge discovery in digital painting datasets, using a deep convolutional neural network for feature extraction and unsupervised nearest neighbor retrieval, with experimental evaluation showing effectiveness on a database of popular artists.
Visual arts are of inestimable importance for the cultural, historic and economic growth of our society. One of the building blocks of most analysis in visual arts is to find similarity relationships among paintings of different artists and painting schools. To help art historians better understand visual arts, this paper presents a framework for visual link retrieval and knowledge discovery in digital painting datasets. Visual link retrieval is accomplished by using a deep convolutional neural network to perform feature extraction and a fully unsupervised nearest neighbor mechanism to retrieve links among digitized paintings. Historical knowledge discovery is achieved by performing a graph analysis that makes it possible to study influences among artists. An experimental evaluation on a database collecting paintings by very popular artists shows the effectiveness of the method. The unsupervised strategy makes the method interesting especially in cases where metadata are scarce, unavailable or difficult to collect.