Hi team,
Here's an issue from the customer may need your help:
Customer used the IWR6843AOP to develop a bathroom fall detection project that required mmWave radar to operate in an indoor bathroom environment with 1.5 m * 1.5 m * 3 m wide height (tile or glass on all sides, mmWave radar on ceiling).
Customers are developing on the basis of the 3D People counting routine, and looking at the point cloud data obtained through Visualizer find the following two issues:
1) When the human body moves rapidly, the same point cloud appears in the symmetrical locations of the walls as the human body, and mmWave radar recognizes it as a human body. That is, when a person stands up and squats in the bathroom, two human bodies are detected.
The customer refer to 3D_people_counting_detection_layer_tuning_guide and 3D_people_counting_tracking_layer_tuning_guide to adjust the parameters by pulling the snrTher of allocationParam to 80 and pointTher to 50, and adjusting the det2actTher in stateParam to 16. The shadow will disappear. Is this adjustment correct? But the shadow is still occasionally detected, what should they do next?
2) Track for positioning the body is always low, more like tracking the lower part of the body. The tracker tracks the upper half of the body based on chest motion only when the person is stationary, but as soon as the lower body moves, Track moves toward the lower body. During this time, the human body still stands, but the radar will indicate that the height has dropped a lot.
The customer is wondering if the configuration of the 3D People counting tracking layer and detection layer can be adjusted to resolve both issues? If not, is there an algorithm similar to eliminating reflections-induced shadow?
Could you help check this case? Thanks.
Best Regards,
Cherry
