People Tracking with the Laplacian Eigenmaps Latent ID: M57 Variable Model. Lu, Carreira-Perpiñán, Sminchisescu
observation Fred latent space tracking result

·

CMU mocap · The Laplacian Eigenmaps Latent Variable Model (LELVM) is a probabilistic dimension reduction method with (A) principled mappings between low-dim latent space and high-dim pose space and (B) densities in the latent and pose spaces. We combine LELVM (trained on mocap data) with particle filters on people tracking tasks. LELVM provides sufficient constraints for robust tracking in the presence of missing, noisy and ambiguous image measurements.