Rotate your phone to navigate through the site properly
Computational and historical Analysis of Hands and Gestures in Early Modern Art
Supervisor : Prof. Dr. Tristan Weddigen
Co-supervisor : Dr. Leonardo Impett
Reproductions, such as replicas of patterns, copies or similar representations of an iconography, have an important place in art history. European artists and their productions, since the early modern times, have been greatly influenced by growing artistic exchanges and the development of reproduction techniques. Furthermore, recent works have shown the possibility to create new browsing tools for art databases with precisely patterns search.
The present work, based on the assumption that we live in a context of visual influences, aims to explore in more details such recurrent patterns and their implications in pictorial art, with a focus on hands. Hand gestures are an important aspect of the narrative systems of paintings in Early Modern Times and can be compared to a specific language that conveys information on the action taking place and the character performing it. Furthermore, the early Renaissance is known to represent an interesting shift in the use of specific hand depictions, from highly symbolic poses to their use in new contexts, more natural movements and other social gestures. Specific codes that do not have been fully decrypted yet.
What are the most recurrent gestures and how do they participate in the narrative system of a painting? Who are the source painters of popular gestures and how to they evolve over time? What are their influences?
As the project is at the intersection of art history and computer science, such primary questions allow to define a good foundation to both types of research.
The elaboration of technological tools to help in engaging with and deepening these research questions is hence clear. The computer science research focus for the time being on automated body pose recognition, technologies that allow to automatically detect and crop hands from a very large collection of digitized artworks and perform further research from an art history perspective.
Based on these machine learning models and computer vision solutions, the present research aims to define a methodology to create and analyse a collection of hands at a large scale. In order to do so, the Fototeca collection of the Bibliotheca Hertziana is used and represents an interesting use case in terms of challenges faced by the field of digital humanities. From accessing the data and metadata to the improvement of the performance of models trained on real images and their further analyses, the project raises questions regarding the value of artificial intelligence tools in the art historical domain and the position of the digital humanist, between the application of new technologies and the real understanding of the core subjects approached.