Demo: Real-Time FMCW Radar Fall Detection with Inception-ResNet 1D (IEEE Study) скачать в хорошем качестве

Demo: Real-Time FMCW Radar Fall Detection with Inception-ResNet 1D (IEEE Study)

📄 Paper Title: Inception-ResNet 1-D CNN With Subject’s Physical Information for Reliable Real-Time FMCW Radar Fall Detection 👨‍💻 Authors: Teguh Patriananda, Istiqomah, Fiky Y. Suratman 📚 Published by: IEEE (Open Access, CC BY 4.0) 🎥 About this Video: This video demonstrates a reliable real-time fall detection system using FMCW millimeter-wave (mmWave) radar integrated with an Inception-ResNet 1-D CNN architecture enhanced by Subject’s Physical Information (SPI). The system is designed to address false alarms caused by fall-like activities (e.g., squatting, bending, sitting) and environmental variations, ensuring stable and accurate real-time deployment. ✨ Highlights: • Privacy-preserving fall detection using FMCW mmWave radar (no camera required). • Deep learning framework based on Inception-ResNet 1-D CNN for temporal modeling. • Integration of Subject’s Physical Information (SPI) including minimum height, vertical displacement, and downward velocity ratio for physics-informed decision validation. • Significant performance improvement:  – Precision increased from 62.71% to 84.90%  – Accuracy improved from 70.27% to 91.11%  – F1-score improved from 77.09% to 91.84%  – Maintained 100% recall under controlled testing conditions. • Robust performance in both controlled and uncontrolled real-world environments. This work demonstrates how combining deep learning with lightweight physics-informed validation can significantly improve reliability and robustness in real-time radar-based fall detection systems for healthcare and elderly monitoring applications. 🔗 For more details, refer to the full article on IEEE Xplore: https://ieeexplore.ieee.org/document/11389751