arXiv subjects

Tiesheng Wang

Publications and source records attributed to Tiesheng Wang.

3 recordsLinked to original sources

arXiv2026

Capturing Uncertainty in Human Motion for Representation Learning in Soccer

This paper presents a self-supervised representation learning framework for understanding 3D skeleton-based human motion in soccer, using future motion prediction as the learning objective. Since human motion is inherently uncertain, accounting for multiple plausible futures is essential for capturing the underlying motion dynamics and learning effective representations. To this end, we introduce a conditioning module for motion prediction that models a probabilistic distribution over discretized future motions in 3D Euclidean space, learning multimodality with explicit supervision from future trajectories. Experiments on large-scale soccer player tracking data show that our approach substantially improves motion prediction accuracy. Moreover, the learned representations effectively transfer to multiple soccer downstream applications, demonstrating strong cross-task generalization.

cs.CV↗

arXiv2025

Unconditional Human Motion and Shape Generation via Balanced Score-Based Diffusion

Recent work has explored a range of model families for human motion generation, including Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), and diffusion-based models. Despite their differences, many methods rely on over-parameterized input features and auxiliary losses to improve empirical results. These strategies should not be strictly necessary for diffusion models to match the human motion distribution. We show that on par with state-of-the-art results in unconditional human motion generation are achievable with a score-based diffusion model using only careful feature-space normalization and analytically derived weightings for the standard L2 score-matching loss, while generating both motion and shape directly, thereby avoiding slow post hoc shape recovery from joints. We build the method step by step, with a clear theoretical motivation for each component, and provide targeted ablations demonstrating the effectiveness of each proposed addition in isolation.

cs.CV↗

arXiv2022

Minimum Surfactant Concentration Required for Inducing Self-shaping of Oil Droplets and Competitive Adsorption Effects

Surfactant choice is key in starting the phenomena of artificial morphogenesis, the bottom-up growth of geometric particles from cooled emulsion droplets, as well as the bottom-up self-assembly of rechargeable microswimmer robots from similar droplets. The choice of surfactant is crucial for the formation of a plastic phase at the oil-water interface, for the kinetics, and for the onset temperature of these processes. But further details are needed to control these processes for bottom-up manufacturing and understand their molecular mechanisms. Still unknown are the minimum concentration of the surfactant necessary to induce the processes, or competing effects in a mixture of surfactants when only one is capable of inducing shapes. Here we systematically study the effect of surfactant nature and concentration on the shape-inducing behaviour of hexadecane-in-water emulsions with both cationic (CTAB) and non-ionic (Tween, Brij) surfactants over up to five orders of magnitude of concentration. The minimum effective concentration is found approximately equal to the critical micelle concentration (CMC), or the solubility limit below the Krafft point of the surfactant. However, the emulsions show low stability at the vicinity of CMC. In a mixed surfactant experiment (Tween 60 and Tween 20), where only one (Tween 60) can induce shapes we elucidate the role of competition at the interface during mixed surfactant adsorption by varying the composition. We find that a lower bound of ~ 75% surface coverage of the shape-inducing surfactant with C14 or longer chain length is necessary for self-shaping to occur. The resulting technique produces a clear visual readout of otherwise difficult to investigate molecular events and establish basic requirements for minimum concentration and protocols to find % surface coverage to induce oil self-shaping by surfactant mixtures.

cond-mat.soft↗