On the Use of Machine Learning and Deep Learning for Radar- Based Passenger Monitoring

In the past decade, there has been an increasing demand for in-vehicle safety sensors. In this paper, we use a multi-input multi-output (MIMO) frequency modulated continuous wave (FMCW) radar for in-vehicle passenger detection and occupant type classification. We propose a Convolutional Long Short-Term Memory (ConvLSTM) that requires no handcrafted rules and accurately detects passengers and classifies occupant type (empty/adults/children). Our model shows high precision (0.90) and recall (0.95) when used to detect unattended children in the vehicles.