Séminaire of the International Research Project (IRP) on Geometric Deep Learning and Generative Models for 3D Human (GeoGen3DHuman)

When: November 26th, 2025, from 9:00am to 12:30pm

Where: UMR CRIStAL, Avenue Henri Poincaré 59655 in Villeneuve d'Ascq, Bâtiment ESPRIT - Amphitheater ATRIUM, ground floor.

Access with Metro M1, station 4 Cantons

Title: Score learning and inference for diffusion processes on shape spaces

Abstract: Steering diffusion processes towards a data distribution is an integral part of diffusion models in generative AI. For geometric data such as shape data, diffusion processes appear as models for stochastic dynamics of e.g. species change through evolution, or for generating data distributions on non-linear spaces, e.g. when defining constructs such as the diffusion mean that relies on geometric equivalents of the Gaussian distribution. Score learning is here key for conditioning on observed data. Thus, score learning provides a connection between generative models and geometric statistics. The talk will concern this connection, bridge simulation on geometric spaces, and application of score learning in geometric contexts. A specific example of this is conditioning diffusion processes in infinite dimensions allowing shape observations to be used for phylogenetic inference in evolutionary biology.

Title: 3D/4D Face and Body Animation

Abstract: Generating dynamic content in 3D is of particular importance for the promise to reduce the cost and development time in many domains like for movie industries, Virtual/Augmented reality, videogames, and humans/avatars interaction. In this talk, we will present some recent results obtained at the Media Integration and Communication Center of the University of Florence in the animation of  synthetic 3D/4D face and body models. First, the task of 3D talking heads generation will be introduced, by discussing the latest advancements that also account for emotions, unconstrained topologies, and style. Then, the task of synthetizing complex behaviors of 3D bodies from text prompts will be addressed, with a specific emphasis on training free solutions, and approaches that also contaminate diffusion models with reinforcement learning in the loop.

Title: Normalizing Diffusion Kernels with Optimal Transport

Abstract: Smoothing a signal based on local neighborhoods is a core operation in machine learning and geometry processing. On well-structured domains such as vector spaces and manifolds, the Laplace operator derived from differential geometry offers a principled approach to smoothing via heat diffusion, with strong theoretical guarantees. However, constructing such Laplacians requires a carefully defined domain structure, which is not always available. Most practitioners thus rely on simple convolution kernels and message-passing layers, which are biased against the boundaries of the domain. We bridge this gap by introducing a broad class of smoothing operators, derived from general similarity or adjacency matrices, and demonstrate that they can be normalized into diffusion-like operators that inherit desirable properties from Laplacians. Our approach relies on a symmetric variant of the Sinkhorn algorithm, which rescales positive smoothing operators to match the structural behavior of heat diffusion. This construction enables Laplacian-like smoothing and processing of irregular data such as point clouds, sparse voxel grids or mixture of Gaussians. We show that the resulting operators not only approximate heat diffusion but also retain spectral information from the Laplacian itself, with applications to shape analysis and matching.

Title: Category agnostic priors for non-rigid shape matching

Abstract: TBA