In response to the widespread health impact of sleep disorders, we present an innovative IoT based sleep detection and alarm system that uses advanced machine learning techniques for real-time monitoring and feedback. This system integrates multi-task cascaded convolutional networks (MTCNN) and multi-column convolutional neural networks with kernelized correlation filters (MCNN-KCF) to provide a comprehensive sleep monitoring solution. MTCNN is used for accurate face recognition and emotional state analysis, providing insight into sleep quality, while MCNN-KCF tracks body movements and sleep stages through video analysis. The system deploys IoT devices, including cameras, microphones, and temperature and humidity sensors, to collect data in the sleep domain. The cameras capture visual data for motion analysis and sleep stage classification, while the MTCNN evaluates facial expressions to detect potential sleep disturbances. This data is transmitted to a central processing unit where machine learning models analyze and interpret the information and offer feedback to improve sleep quality. The goal of this approach is to provide a discreet, cost-effective solution for monitoring and improving sleep health.