This research study aims to create a smart child activity recognition device combines a Raspberry Pi-pico, sensors for Internet of things (IoT). The system includes several sensors for monitoring human motion, pulse, and location parameters. Different types of machine learning and deep learning algorithm is used for analyzing data to identify various kinds of activities in children including sitting, standing, running, and climbing the stairs. This study utilizes a Raspberry Pi-pico, which is a cost-effective as well as tiny single-board computer central processing unit for an infant monitoring system. It can be easily integrated with various sensors and devices, which make it a suitable foundation for developing an IoT-based tracking system. It is able to link to the Internet through Wi-Fi, allowing for surveillance remotely. The dataset collected from 22 different children based on their features like sitting, standing, running, and climbing the stairs. The model was trained using a data of child mobility and behavior, distinguishing activities such as sitting, standing, running, and stair climbing. There are six different classifiers: LSTM, SVM, ANN, KNN, AdaBoost, and Logistic Regression. When applied to pre-processed data, SVM outperformed other classification methods. The classification accuracy reached 98.81%. The experimental results indicate that the proposed system may identify behavior and it will keep parents remotely. The technology continuously monitors and analyzes the child’s environment and behavior to ensure their safety and well-being.