Empowering Communication: Utilizing Facial Expressions to Classify Interruptibility in the Workplace

Abstract In today's information-driven workplaces, interruptions have grown commonplace and substantially negatively influence individual and team performance. To create a productive work environment, it is essential to comprehend the connection between interruptions and stress. This research paper focuses on utilizing facial expressions to gauge people's stress levels while examining the connections between interruptions, stress, and the consequences of positive or negative feedback. Facial expression analysis may provide important clues about someone's emotional condition, helping us choose when and how to interrupt them. Through the examination of facial expressions, this integrated method seeks to create a holistic system that can reliably identify stress levels while fostering respectful and effective communication techniques. The Interruptibility classification deep convolutional neural network architecture is based on pretrained models (Sequential model) using the Facial Expression Recognition Plus (FERPlus) dataset with Haar Cascade face detection. The small weight of the Sequential model and its better latency performance are its main benefits. Updated annotations are included in the FERPlus dataset for precise emotion categorization. This research helps designers of interruptibility classification systems create more precise systems by revealing patterns and connections between facial expression metrics and stress or non-stress states, promoting a positive and effective work environment. The results provide useful recommendations for controlling interruptions depending on people's stress levels, promoting efficient workplace communication, and minimizing unnecessary disturbances.