par Cakoni, Dejvi
;Storrer, Laurent
;De Doncker, Philippe
;Horlin, François ![](/vufind/images/ULB/publications_list.png)
Référence (06-10 November, 2023: Sydney, Australia), Proc. of the 2023 IEEE International Radar Conference (RADAR)
Publication Publié, 2023-11-06
![](/vufind/images/ULB/publications_list.png)
![](/vufind/images/ULB/publications_list.png)
![](/vufind/images/ULB/publications_list.png)
![](/vufind/images/ULB/publications_list.png)
Référence (06-10 November, 2023: Sydney, Australia), Proc. of the 2023 IEEE International Radar Conference (RADAR)
Publication Publié, 2023-11-06
Publication dans des actes
Résumé : | People counting and detection technologies have shown great versatility in various scenarios and have become an important tool for event organizers and city planners to optimize their operations. This paper presents a novel approach for people counting using Micro-Doppler Signatures (MDS) extracted from a Frequency-Modulated Continuous-Wave (FMCW) radar operating at 77GHz. The system utilizes the unique gait model of each individual, which results in a distinct instantaneous velocity over time, to generate the MDS that are later used to classify groups of different sizes with a Convolutional Neural Network (CNN). Those results are compared with using existing CNNs for image classification, in a transferred learning approach. The proposed system overcomes the limitations of existing camera-based people counting techniques such as the need for a clear line of sight and being affected by lighting conditions. |