Abstract: “VisionSense” is an AI-powered facial recognition system that performs real-time face identity, age, gender, and emotion prediction. It uses a Convolutional Neural Network (CNN) with pre-trained weights for feature extraction. It uses a pre-trained dlib model for face alignment. It uses a pre-trained Wide Residual Network model for age and gender predictions. It uses another Convolutional Neural Network (CNN) for emotion detection. For identity classification, the system offers a choice between K-Nearest Neighbours (KNN) and Support Vector Classification (SVC). VisionSense is designed to process images, videos, and live camera feeds, automatically saving unknown faces and updating its model through retraining whenever a new customer is detected. The system provides a web-based API for seamless image and video input, allowing users to interact with it effortlessly.

Keywords: Convolutional Neural Network, Wide Residual Network, K-Nearest Neighbours, Support Vector Classification, Machine learning, Face classification, Emotion detection

| DOI: 10.17148/IARJSET.2025.12405