<![CDATA[With the increasing internet usage post-pandemic, ensuring the security of a fintech application becomes imperative. Bangbeli implements KYC procedures using facial recognition technology and stringent security protocols to verify identities and safeguard users' personal data in compliance with Bank Indonesia regulations. Utilizing Haar Cascade Classifier, Local Binary Pattern Histogram, and histogram equalization, an API (Application Programming Interface) has been created for facial training and prediction. These methods were chosen for their credibility, achieving an 88% accuracy with 33 samples and 90% with 10 samples. This study focuses on constructing an API for mobile services at Bangbeli, achieving 87.5% accuracy, 81.25% precision, 87.5% recall, and a 25% error rate. The model demonstrates good performance in facial recognition, with an acceptable error rate. Although precision is slightly lower than recall, it suggests the model is more inclined to identify most positive data with some errors rather than discard potentially identifiable faces.]]>
Copyrights © 2024
