This paper presents a computer vision color tracking algorithm, specifically the Continuously Adaptive Mean Shift (CAMSHIFT) algorithm, designed for tracking human faces in real-time applications. The goal is to develop a perceptual user interface that can track and understand human motion, pose, and tool use. CAMSHIFT is based on the mean shift algorithm, which is a robust non-parametric technique for finding the mode of probability distributions. The modified CAMSHIFT algorithm dynamically adjusts its search window size to track color probability distributions derived from video frame sequences. The paper compares CAMSHIFT's tracking accuracy with a Polhemus tracker and evaluates its performance in noisy environments and the presence of distractors. CAMSHIFT is also used to control commercial computer games and explore immersive 3D graphic worlds. The algorithm's simplicity and computational efficiency make it suitable for real-time applications, and its robustness to noise, distractors, and occlusions enhances its practical value. The paper concludes by discussing the limitations and future potential of CAMSHIFT in more complex tracking systems.This paper presents a computer vision color tracking algorithm, specifically the Continuously Adaptive Mean Shift (CAMSHIFT) algorithm, designed for tracking human faces in real-time applications. The goal is to develop a perceptual user interface that can track and understand human motion, pose, and tool use. CAMSHIFT is based on the mean shift algorithm, which is a robust non-parametric technique for finding the mode of probability distributions. The modified CAMSHIFT algorithm dynamically adjusts its search window size to track color probability distributions derived from video frame sequences. The paper compares CAMSHIFT's tracking accuracy with a Polhemus tracker and evaluates its performance in noisy environments and the presence of distractors. CAMSHIFT is also used to control commercial computer games and explore immersive 3D graphic worlds. The algorithm's simplicity and computational efficiency make it suitable for real-time applications, and its robustness to noise, distractors, and occlusions enhances its practical value. The paper concludes by discussing the limitations and future potential of CAMSHIFT in more complex tracking systems.