Lucas Hirsch

Lucas.Hirsch[at]ed[.]ac[.]uk

Edinburgh, UK

Buenos Aires, Argentina

I am a PhD researcher in machine learning and signal processing at the University of Edinburgh, supervised by Mike Davies. My research focuses on representation learning for sensing and imaging, particularly self-supervised and cross-modal methods that work with limited labelled data.

Previously, I worked as an R&D Engineer at Tenaris developing optical non-destructive testing systems. Before that, I researched photoacoustic tomography and inverse problems at the University of Buenos Aires.

You can find my work on my publications page or Google Scholar.

Outside research, I enjoy hiking, solving crosswords, playing bass and collecting random facts and slightly weird ideas. My favorite scientists include Claude Shannon and Oliver Heaviside.

Feel free to reach out to discuss research collaborations, ideas or anything that you think might interest me!

news

Sep 01, 2026	“Cross-Modal Knowledge Transfer for SAR Target Recognition Using Optical Vision Foundation Models” was accepted for presentation at SPIE Sensors + Imaging 2026! The work explores how optical vision foundation models can support SAR target classification without requiring optical imagery at inference time. I’ll be presenting it as a poster on 16 September 2026. Paper available on arXiv.
Jun 20, 2026	The paper including our solution (Team IDCOM) for the 5th Multi-modal Aerial View Imagery Challenge: Classification (MAVIC-C) has been published! Code available on GitHub.
Mar 03, 2026	I’ve placed 3rd overall (and 1st in OOD detection) in the 5th Multi-modal Aerial View Imagery Challenge: Classification (MAVIC-C)! A technical report detailing the solution will appear in the Perception Beyond the Visible Spectrum workshop (PBVS) at CVPR 2026. Code available on GitHub.
Jul 01, 2025	I’ve started my PhD at the University of Edinburgh!

selected publications

Cross-Modal Learning for SAR Target Recognition Using Optical Vision Foundation Models

Lucas Hirsch, James R. Hopgood, Javid Khan, and 2 more authors

Sep 2026

arXiv PDF Code
On the robustness of model-based algorithms for photoacoustic tomography: Comparison between time and frequency domains

Lucas Hirsch, M. G. González, and L. Rey Vega

Review of Scientific Instruments, Nov 2021

Abs DOI arXiv PDF

For photoacoustic image reconstruction, certain parameters such as sensor positions and speed of sound have a major impact on the reconstruction process and must be carefully determined before data acquisition. Uncertainties in these parameters can lead to errors produced by a modeling mismatch, hindering the reconstruction process and severely affecting the resulting image quality. Therefore, in this work, we study how modeling errors arising from uncertainty in sensor locations affect the images obtained by matrix model-based reconstruction algorithms based on time domain and frequency domain models of the photoacoustic problem. The effects on the reconstruction performance with respect to the uncertainty in the knowledge of the sensors location are compared and analyzed both in a qualitative and quantitative fashion for both time and frequency models. Ultimately, our study shows that the frequency domain approach is more sensitive to this kind of modeling errors. These conclusions are supported by numerical experiments and a theoretical sensitivity analysis of the mathematical operator for the direct problem.