International Journal of Science, Engineering and TechnologyAuthors: Yash Solanki, Heerav Amin, Diya Chunara, Dhvani Puwar, Prof. Yatin Shukla
Abstract: Artificial Intelligence (AI) is increasingly transform- ing the way humans interact with digital systems by enabling emotionally responsive, empathetic companions. The AI Friend Web App is designed as an interactive assistant capable of detecting real-time emotions and identifying surrounding objects using integrated camera feeds. Unlike conventional chatbots that lack contextual sensitivity, this system integrates multimodal emotion detection (facial and vocal analysis), conversational AI powered by Google Gemini, and expressive 3D avatars to deliver a more human-like experience. The system ensures secure handling of sensitive user data through Firebase integration. This paper presents the architecture, methodology, performance evaluation, and future scope of the AI Friend project. Results demonstrate emotion detection accuracy above 85%, object detection latency under 300 ms, and over 90% contextual relevance in chatbot responses. User feedback highlights significant improvements in emotional engagement and perceived empathy. These findings underline the potential of empathetic AI systems to contribute meaningfully to mental health support, human-computer rela- tionships, and adaptive companionship technologies.
DOI: http://doi.org/10.5281/zenodo.17598876
