ThermalGait

Crowded public spaces present a significant security challenge where conventional surveillance systems struggle to accurately identify potential threats without producing a high number of false positives. Manual screening is often intrusive, time-consuming, and difficult to scale in real-world environments.

Designed a multi-stage threat detection pipeline that combines gait recognition, human activity analysis, and thermal imaging. The system progressively filters suspicious individuals using behavioral analysis before validating potential threats through thermal screening, improving both efficiency and detection confidence.

Developed a privacy-conscious security framework capable of improving concealed threat detection while reducing false positives. The project demonstrated how multi-modal AI techniques can enhance public safety without relying on intrusive surveillance methods.