Currently, there is a growing interest in Smart home systems using the
Internet of Things. One of the most important aspects of the Smart home
system is the security capability of being able to easily lock and unlock doors
or gates. The main problem in smart home security systems is the low accuracy
and image processing delay of around 70% -65% in the experiments studied
using the KNN and Decision Tree methods. . This proposal proposes the Deep
Learning method which has an accuracy of 80% or more. The methods used
in this final project research are 1. Literature study on SmartHome Security,
2. Perform analysis and performance, 3. Test Algorithms for CNN, DNN, and
RNN. It is hoped that the proposed Deep Learning method can obtain facial
recognition accuracy above 80%.
Keywords: Smart Home Security ; KNN; Deep Learning;
RNN;CNN;Decision Tree.