Leonardo Impett

Leonardo Impett



Research Assistant


Image and Visual Representation Lab and Digital Humanities Institute

Research Interests

The bulk of my research, and my PhD, is at the intersection of computer vision and art history – supervised by Sabine Suesstrunk (IVRL) and Franco Moretti (Digital Humanities). 

Research Projects

Aby Warburg’s Bilderatlas and the Pathosformel

Aby Warburg understood that art, passions and emotions are not only portrayed mimetically. Sometimes, they come to us through repeated formulas – most often, formulas of the body – which he christened Pathosformeln. These Pathosformeln move from antiquity to the modern period, in mythological or everyday scenes, from high art to popular book prints. Neither depiction nor symbol, Pathosformeln lie in the Zwischenraum that Warburg identified between semiotic and pritive reactions. His most ambitious project, the Bilderatlas, was unfinished by his death.


My collaborative project with Franco Moretti tries to understand the Pathosformel and its role in the Bilderatlas through computational and formal means. We attempt to operationalise the Pathosformel – that is, to turn it into a sequence of quantitative operations – and in doing so reach a better critical understanding of the Pathosformel itself, as in the disassembling of a clock-mechanism. By reducing the Pathosformel to its bare constituents, viewing them as statistical phenomena, we find a remarkable unity across Warburg’s taxonomy of passion.


Computer Vision in Paintings

In order to do computational, content-based art history with computer vision, we need to know how to build computer vision systems that work on paintings, bas-reliefs, scultpures, and so on. Most – if not all – of the research focus in computer vision has, however, been on photos. Instead of starting from scratch again, I’m interested in trying to use networks, datasets and techniques from photo-based computer vision in paintings. This isn’t just a case of domain transfer, I claim, but of generalisation.


I focus on two problems: pose detection in paintings (for detecting Pathosformeln automatically), and generalisation of photo-based computer-vision models to paintings (including style tranfser). 


Gesture in Art, Gesture in Hieroglyphs, Gesture in Psycholinguistics

Psycholinguists such as Adam Kendon and David McNiel propose a great deal of unity between what was previously considered (spoken) language and gesture. Together with Alexandre Loktionov at the Library of Congress, we look at gesture which is both static and purposefully depicted – such as a painting or hieroglyph – seeing how linguistic theory (normally based on ‘live’ gestures) translates (or doesn’t). The examples of Pathosformeln . Furthermore, linguistic models for the development of gesture – combined with Warburg’s theory of Pathosformeln – could form the basis of a formal analysis of pose from paintings, automatically estimated by computer-vision. 


Ways of Machine Seeing

John Berger popularised the ideas of Walter Benjamin in his 1972 BBC TV series, Ways of Seeing, where he showed how the way we see art was influenced by material (class, patron, profit, advertising) and technological (photography, reproduction, the ‘detail’ postcard) conditions. The Ways of Seeing group is formed by me, Prof. Alan Blackwell (Cam), Dr. Anne Alexander (Cam) and Prof. Geoff Cox (Aarhus). We seek to update this analysis to computer vision, and more generally, to machine ways of seeing. We organise a yearly workshop in Cambridge called Ways of Machine Seeing. 


Artistic Research and Critical Technical Practice

It seems to me that the Digital Humanities, Critical Technical Practice (Philip Agre’s strand of artificial intelligence research) and Artistic Research are the three principal meeting-points of critical theory and technical research – yet they seldom talk to each other. To this end, I’m organising the first ETH-domain summer school on CTP and AR in the digital humanities. Link soon – watch this space – but for more information, contact me directly. 

Student Projects


Current Students

Yalan Yue, Exploring the Bilderatlas with Augmented Reality

Kim Phan, Interactive browser-based exploration of art historical research

Bruno Wicht, Non-photorealistic rendering of human pose and shape

Emma Clement, Deep learning for emotional image classification and enhancement, University of Cambridge


Open projects for Autumn 2017: 


Individual student proposals are welcome – please get in touch well before the deadline to arrange. 


Summer internships should refer to the Summer@EPFL program for more details. 

Other Affiliations


Fellow of the Royal Society of the Arts, London

Member of the Institution of Engineering and Technology

External Member of the Rainbow Research Group, University of Cambridge



Waking the Monsters of Insomniac Rationality: Conspiracy theory as Critical Technical Practice (2017)

Impett, L., Gibson, A., Djurslev, A. & Blackwell, A. Chapter to appear in The Monster and The Monstrous, Berlin (accepted)



From Mnemosyne To Terpsichore – The Bilderatlas After The Image (2017).  Impett, L. & Süsstrunk, S., To appear in Digital Humanities 2017 (accepted)


Pose and Pathosformel in Aby Warburg’s Bilderatlas (2016). Impett, L. & Süsstrunk, S. In VISART, European Conference on Computer Vision.


Vygotsky and Marxist artificial intelligence (2015). Impett, L. Theory and Struggle, Marx Memorial Library, London.


Automatic face analysis tools for interactive digital games (2015). Mahmoud, M., Baltrusaitis, T., Gavrila, V., Rozycki, M., Impett, L. & Robinson, P. Intelligent Digital Games for Empowerment and Inclusion 2015


Augmenting Bioacoustic Cognition with Tangible User Interfaces (2015). Herman, I., Impett, L., Wollner P.K.A. & Blackwell, A.F.  HCI International 2015


The Mephistophone: Exploring Musician Fantasies of Traditional-Instrument Interactivity (2015).   Impett, L., Herman, I., Wollner, P.K.A. & Blackwell, A.F. HCI International 2015


Mephistophone (2014).  Wollner, P.K.A., Herman, I., Pribadi, H., Impett, L., and Blackwell, A.F. University of Cambridge Computer Laboratory Technical Report UCAM-CL-TR-855.


A Facial Affect Mapping Engine (2014).  Impett, L., Robinson, P., & Baltrusaitis, T. ACM Conference on Intelligent User Interfaces 2014, Haifa, Israel.


A Facial Affect Mapping Engine (2014). Impett, L., Robinson, P., & Baltrusaitis, T. Intelligent Digital Games for Empowerment and Inclusion, Haifa, Israel.


The perception of potential: interference, dimensionality and knowledge (2013). Impett, L., & Impett, J.

Advances in Artificial Life ECAL 2013. Proceedings of the twelfth European Conference on the Synthesis and Simulation of Living Systems. MIT Press, pp. 1126-1131.