AI Technology: Vision (image recognition, image generation, text recognition OCR, etc.) | Deep Learning
This project has been awarded by the Fortissimo FFPlus funding call for Innovation Studies (Proposal id: 319), and the access to EuroHPC is being requested following recommended steps by the Fortissimo consortium. Skeleton-based forensic human identification strongly relies on manual, subjective and error-prone methods that can largely benefit from data-driven automated software alternatives. In particular, there is a demand for a software solution that can recreate the face of a person, often an unidentified cadaver, from the information that can be extracted from the structure of the skull. This process is called CranioFacial Reconstruction (CFR), a technique that nowadays requires very specific expertise of forensic artist experts that provide a service that is costly, slow and with limited access. An accurate CFR software tool would be a game-changer for forensic casework for cases where no other identification method is viable, mainly due to the lack of ante mortem data, thus remaining unsolved and going cold while grieving families await news about their missing loved ones.
The advent of generative Artificial Intelligence (AI) finally makes an automated CFR tool possible. For Panacea, as an AI software SME solely focused on the forensic field, it would mean achieving a unique position in the market with an unmatched technology that would create new business models, new revenue streams and improve the accuracy of an ID method already in the market. The present work plan aims to define the best methodology to achieve a functional CFR model comparing different strategies that generative AI allows for. Notably, the recent release of open-source foundation models constitutes a very promising approach to CFR, that could provide an unprecedented level of realism in the results. Nonetheless, such approaches are very costly in computational resources and the development of this project with its many steps, even if executed in parallel, becomes unfeasible owing to impractical execution times. Thus, the support of large-scale High-Performance Computing (HPC) services together with the right expertise will be critical for its success.
For this project, experts in generative AI for images and forensics from the University of Granada (UGR) and the National Institute for Research in Digital Science and Technology (INRIA), will complete Panacea’s AI expertise precisely covering the most critical knowledge necessary for this development. The present consortium is guaranteed to make the most of FACEGEN and enhance the knowledge transfer from the academic realm to the market, and thus, to the forensic community and the public at large.
Andrea Valsecchi, Panacea Cooperative Research, Spain