Design of Elderly Gaming Assistance Framework through ContextAware Object Recognition and Immersive Interaction

The rise in the number of elderly individuals necessitates the development of interactive systems that facilitate safe exercise and mental stimulation. This paper proposes the Elder Vision-GA, which is an adaptive assistive gaming framework that leverages real-time object detection and personalized gameplay experience. The proposed approach uses a lightweight convolutional neural network architecture alongside an attention-based feature enhancement layer to optimize detection performance of indoor objects regardless of changes in illumination and surroundings. A cognitive load assessment model is used to estimate user interaction efficiency through response time, action deviations, and hesitation behaviour. Such features are integrated into a reinforcement learning-based adaptation system that dynamically varies object scale, motion velocity, visual appearance, and difficulty level. Furthermore, a safety monitor unit detects any instability and risk of fall during interaction. Experiments reveal promising results in terms of object detection performance, efficient real-time operation on embedded devices, and task completion rate and user interaction efficiency over non-adaptive systems