Convolutional Neural Network for Hand Gesture Detection in an IoT-Based Smart Lock System

In the modern era, security is a major concern, with many cases of theft involving traditional locks. This research aims to develop an IoT-based smart lock system that can be controlled by an android application. The system uses Convolutional Neural Network (CNN) for hand gesture recognition as the control method and TensorFlow Lite for inference on mobile devices. Assessment showed an average accuracy of 96.80% for the closed hand gesture (closing the door) and 96.27% for the open hand gesture (opening the door), with a response time of 0.1 seconds. The system improves efficiency and security and provides easy remote access. Despite challenges such as gesture recognition in low-light conditions, the system provides an innovative solution for improved facility security.