July 3, 2023

Research on milli-wave data processing for in-cabin passenger detection presented at MIRU 2023

Research on milli-wave data processing for in-cabin passenger detection presented at MIRU 2023
July 2023, our paper entitled "Deep Neural Networks for in-cavin passenger recognition by feeding raw data from milli-wave sensor" was presented at Symposium on Image Recognition and Understanding (MIRU) 2023.

Although the milli-wave data is noisy especially in closed space, our approach can detect passenger existence better than conventional methods which make a distance map. In radar (milli-wave data) processing, most of the methods make the distance map by using Fast Fourier Transform (FFT) as a preprocessing. It can extract angle, distance and speed of object, however, it removes details of original signals which might include important information for object detection. In contrast, our method feeds original raw data, whose form is complex number, directory to DNN and performs object detection in order to make the most of details of the signals. Thanks to this approach, we achieved better results than conventional methods.