SmartPosture: Real-Time Posture Detection and Feedback System Integrating IoT and Machine Learning

Traditional posture monitoring methods often rely on physical correction, lacking advanced technologies for real-time detection and feedback, which can result in delayed interventions. This research presents a machine learning framework designed to address this deficiency by giving instantaneous posture detection and feedback. The framework categorizes user posture into “Good” or “Bad” classifications, achieving an accuracy rate of 75.52%, by the use of Convolutional Neural Networks (CNNs) along with TensorFlow. Trained on a 2,000-image Yoga Poses Dataset and fine-tuned over 50 epochs, the model achieved optimal performance. The system integrates IoT and machine learning, providing recommendations for posture -related advice through a medical chatbot powered by techniques of NLP like Langchain and Hugging Face embeddings. Key hardware components include the flexible sensor, LCD display, vibration motor, and ESP8266 Wi-Fi module, which offer real-time alerts and monitoring. The web application, built using Flask and Streamlit, allows users to track posture data and receive corrective prompts. This innovative approach not only enhances posture monitoring but also advances smart healthcare, offering a continuously learning and adaptable solution relevant to real-world scenarios.