A Vision-Based AI Framework for Real-Time Fatigue and Workload Detection in IT Professionals Using MediaPipe and a Fusion Neural Network

In the prevailing work setting of the modern technology sector, screen usage, static positions, and cognitive engagements of the brain contribute to physical and mental exhaustion. Existing solutions to fatigue monitoring and alerting are often computationally complex and wearable and invasive technology. This research work introduces the use of a vision-tracking AI model that is non-invasive and exclusive to the specific requirements of the technical professionals. The model considers the eye movements, body positions, and human interactions to provide an accurate level of physical and mental fatigue. Through the learning concept of fusion learning, the model differentiates between the drastic and short-lived work patterns and the continuous physical and mental states. The proposed model is validated to collectively work in a timely and expert manner with very low computational complexity, thereby imparting expert warning notifications related to physical and mental fatigue. The model adheres to the concepts and requirements of Industry 5.0.