Counting people and knowing the density or average distance between people is important information for many different applications. Especially now, due to the current Covid-19 situation, we have a high demand for a device that can accurately count people in indoor environments such as cinemas, lecture rooms at universities, public transportation, etc. The device should be very simple to install, battery powered without any cables to connect, and all measurements and processing should be performed inside the device.In this proposal, we will investigate the feasibility of abattery-powered edge-computing device that provides data-centric processing in the mW range. It should use the new RISC-V processor to perform machine learning algorithms (neural networks) on images captured with an ULP camera (Ultra-Low Power Camera). The RISC-V processor is well suited for low-power performance. The results of the image analysis are transmitted using a low-power communication technology (LoRa and LoRaWAN®) to further improve the battery life of the sensor device. In addition, personal data and privacy is protected because the entire data processing is performed in theembedded device and no images are transmitted out of the device. In case of a successful feasibility, two PoCs are planned to be installed at:-Zürcher Hochschule für Angewandte Wissenschaften for meeting and lecture room monitoring-SBB for railways monitoring