Iot based smart home entrance automation using Raspberry Pi

V zaključni nalogi s pomočjo platforme Raspberry Pi razvijemo in implementiramo lasten avtomatiziran sistem za vstop v dom. Osnovne dogodke zaznamo s pomočjo kamere in algoritma zaznave gibanja. Primerjamo in testiramo dve metodi zaznave gibanja na osnovi modela Gaussovih mešanic in Bayesove verjetnosti. Vstop v dom poteka na osnovi dvostopenjske avtentikacije, kjer se preverjata biometrična in brezkontaktna identifikacija. Biometrično identifikacijo izpolnimo že pri 80 % gotovosti prepoznave obraza, brezkontaktno pa s pomočjo zaznane naprave, ki pripada določenemu profilu. Z uporabo tehnologije interneta stvari avtomatiziran sistem samodejno upravlja z lučmi v prostoru glede na sončni vzhod in zahod, predvaja glasbo in regulira temperaturo zraka glede na letni čas. Celotni sistem testiramo 30 dni, kjer je bila odzivnost sistema manjša kot 5 sekund in uspešnost odločanja sistema večja kot 90 %. Uporabo sistema dodatno podpremo z enostavno namizno aplikacijo, ki uporabniku omogoča ogled video prenosa v živo, konfiguracijo osebnih profilov, ročno upravljanje vhodnih vrat in pregled zgodovine dogodkov.The thesis describes the development and implementation of an automated home entrance system, based on the Raspberry Pi platform. System primary events are detected by a camera and a motion detection algorithm. We compare and test two motion detection methods based on the Gaussian mixture model and the Bayesian probability. Home entrance is based on two-factor authentication where biometric and contactless identification is verified. Authentication is fulfilled with biometric identification in the case of 80 % certainty of face recognition and with the help of a contactless detected device that belongs to a specific profile. With the support of Internet of Things technology, the automated system automatically controls the lights in the room concerning sunrise and sunset, plays music, and regulates the air temperature according to the time of year. We tested the entire system for 30 days, where the system\u27s responsiveness was lower than 5 seconds, and the decision-making efficiency of the system was greater than 90 %. The use of the system is supported with the desktop application that allows the user to watch live video streaming, configure personal profiles, manually operate the front door, and view event history