We present an evolutionary adaptive eye-tracking framework aiming for low-cost human computer interaction. The main focus is to guarantee eye-tracking performance without using high-cost devices and strongly controlled situations. The performance optimization of eye tracking is formulated into the dynamic control problem of deciding on an eye tracking algorithm structure and associated thresholds/parameters, where the dynamic control space is denoted by genotype and phenotype spaces. The evolutionary algorithm is responsible for exploring the genotype control space, and the reinforcement learning algorithm organizes the evolved genotype into a reactive phenotype. The evolutionary algorithm encodes an eye-tracking scheme as a genetic code based on image variation analysis. Then, the reinforcement learning algorithm defines internal states in a phenotype control space limited by the perceived genetic code and carries out interactive adaptations. The proposed method can achieve optimal performance by compromising the difficulty in the real-time performance of the evolutionary algorithm and the drawback of the huge search space of the reinforcement learning algorithm. Extensive experiments were carried out using webcam image sequences and yielded very encouraging results. The framework can be readily applied to other low-cost vision-based human computer interactions in solving their intrinsic brittleness in unstable operational environments.

This paper presents an evolutionary and adaptive framework for efficient visual tracking based on a hybrid POMDP formulation. The main focus is to guarantee visual tracking performance under varying environments without strongly-controlled situations or high-cost devices. The performance optimization is formulated as a dynamic adaptation of the system control parameters, i.e., threshold and adjusting parameters in a visual tracking algorithm, based on the hybrid of offline and online POMDPs. The hybrid POMDP allows the agent to construct world-belief models under uncertain environments in offline, and focus on the optimization of the system control parameters over the current world model in real-time. Since the visual tracking should satisfy strict real-time constraints, we restrict our attention to simpler and faster approaches instead of exploring the belief space of each world model directly. The hybrid POMDP is thus solved by an evolutionary adaptive framework employing the GA (Genetic Algorithm) and real-time Q-learning approaches in the optimally reachable genotype and phenotype spaces, respectively. Experiments were carried out extensively in the area of eye tracking using videos of various structures and qualities, and yielded very encouraging results. The framework can achieve an optimal performance by balancing the tracking accuracy and realtime constraints.

No abstract