Smart Camera Traffic Behaviour Classification: A Detailed Investigation Using Machine Learning

The rise of the Internet of Things (IoT) has revolutionized how people interact with their surroundings. To address concerns about flexibility and effectiveness in behavior detection, we implement machine learning (ML) algorithms to classify network activity. We used Wireshark to collect smart camera traffic data and tested ML models including RF, KNN, and SVM. These were combined into an ensemble model that achieved 99% accuracy in classifying attack types and devices. SHAP (Shapley Additive Explanations) was used to interpret how features contributed to model decisions. These results show that ML models, combined with interpretability tools, offer effective and explainable cybersecurity solutions for IoT systems.