EmotiSense AI

EmotiSense AI

V.Adithya

Department of Computer Science and Engineering

Presidency University Bengaluru, India adith348@gmail.com

Pakruddin B

School of CSE & IS Presidency University

Bengaluru, India fakrubasha@gmail.com

Abstract— Simply, the EmotiSense AI project enhances human-computer emotional interaction using an accessible, cross-platform emotion recognition system. In the paper, a multimodal AI system will be developed using facial cues, vocal signals, and textual inputs to detect and interpret emotions in

real time. This facilitates meaningful, context-aware interactions between users and machines using convolutional neural networks (CNNs) and BiLSTM models, coupled with the DeepFace and RoBERTa language model. The integrated system is designed to work offline, dynamically generating task recommendations tailored to the user's emotional state. The emotion-aware assistant is particularly designed to support individuals with autism and depression by offering personalized, empathetic task suggestions and maintaining emotional well-being. Discussions also include implementation challenges of cross-modal synchronization, the ethics of affective computing, and the accessibility impact of such intelligent assistants. Results are interactive, low-latency, and effective, demonstrating high user engagement and satisfaction in real-time emotion analysis and support.

Index Terms—Emotion Detection, Multimodal AI, DeepFace, RoBERTa, Affective Computing, Mental Health, Real-time Analysis