MARVEL’s first scientific publication focuses on “Enabling energy efficient machine learning on a Ultra-Low-Power vision sensor for IoT”. The paper presents a way to extract meaningful information from cameras installed in a smart city that usually have to trade image quality for energy efficiency. This is a typical bottleneck in the current technologies used for providing real-time insights, in order to deliver useful services to the citizens. The paper presents a pipeline, demonstrating how it is possible to get good quality information from low-quality images.

(Featured photo by JC Gellidon on Unsplash)