Pfinder demonstrates the utility of stochastic, region-based features for real-time image understanding. This approach allows meaningful, interactive-rate interpretation of the the human form without custom hardware. The technique is stable enough to support real applications as well as higher-order vision techniques.