ANTI-THEFT MODEL FOR SMART HOMES

ABSTRACT

Smart homes enable remote control of appliances and increase user convenience, but a major challenge persists in detecting intruders with masked or covered faces. Existing systems, although effective under normal conditions, are not accurate when faces are obscured. The aim of this research is to develop an anti-theft model that accurately detects intruders even with face coverings. Augmentation techniques expanded the dataset to 90,960 images, ensuring diversity. The CNN model was trained on this dataset and evaluated using accuracy, precision, recall, and F1-score, all of which returned a perfect score of 1.00. The results demonstrate that the proposed CNN-based system can reliably distinguish between homeowners and masked intruders in real time. This solution addresses a critical limitation in existing facial recognition systems and enhances the overall effectiveness of smart home security. making smart homes safer.

Keywords: Anti-theft, CNN, face recognition, intruder detection, Flask, real-time alert