The ACLab operates at the forefront of computer vision, machine learning, and pattern recognition, with a mission to understand and predict human and animal behaviors through visual perception tasks such as object modeling, tracking, and action recognition. By transforming complex data into low-dimensional “pose” embeddings, the lab achieves efficient yet interpretable state estimation, harnessing machine intelligence to augment—rather than replace—human capabilities.

A key focus of the lab is addressing Small Data challenges by integrating domain knowledge into generative models and developing data-efficient machine learning approaches. These efforts are particularly impactful in critical domains where data collection and labeling are inherently challenging, costly, or constrained by privacy and classification laws. Unlike conventional ML and computer vision models that rely on large datasets, ACLab’s work is tailored for Small Data contexts, which dominate real-world applications. Through innovative techniques, the lab overcomes the scarcity of labeled data and high acquisition costs, driving progress in scientific and industrial fields where traditional data-intensive methods fall short.