Investigating and Deploying an AI model onRaspberry Pi IoT platform using FIWARE andDocker in the context of the Bonseyes AIEcosystem

Sammanfattning: Many high-end sophisticated devices are being replaced by small single-board IoT devices. Itis estimated that by the year 2025 there will be about 75 billion IoT connected devicesworldwide. This project buttresses the earlier statements. The project will be focused on howto deploy a complex AI face-recognition model on a simple, less power consumption, single-board IoT device Raspberry Pi3. Also, to simplify the whole process of deploying the AI modelon Raspberry Pi3 by dockerizing. By just pulling the respective docker image and running it onthe end device (Raspberry Pi3), we can make face recognition happen. Finally, the obtainedresults will be sent to the cloud (Fiware) using NGSI API. Instead of sending the whole videofeed to the Cloud platform and performing the computing there, we managed to perform thecomputing at the edge and then send the results to the cloud and making the results accessibleto requested users.