This project explores using Wi-Fi signals to detect human presence and estimate their poses in an indoor environment, without cameras or wearables. The research aims to characterize the impact of human pose on Wi-Fi signals and develop deep-learning models to map 1D signals to 2D pose. A dataset of Wi-Fi channel state information (CSI) from 4 volunteers is used to train a deep-learning model, achieving 60.39 % accuracy on CSI data. The system allows contactless, privacy-preserving human sensing for applications like rescue operations, military applications, and elderly monitoring, leveraging Wi-Fi infrastructure beyond communication. Field tests validate the system's performance in an indoor environment, demonstrating the potential of Wi-Fi-based vision-free human sensing.
@artical{d1352024ijcatr13051001,
Title = "Human Detection And Pose Estimation Using Wi-Fi Signals",
Journal ="International Journal of Computer Applications Technology and Research (IJCATR)",
Volume = "13",
Issue ="5",
Pages ="1 - 7",
Year = "2024",
Authors ="Dr. S Jagadeesha, Dr. Prathibha Kiran, Yaramasa Gautham, Sindhu R, Shaan Ghosh, Sagar N Sankanatti"}