The manual human resource management (HRM) system at BMT Bahtera
Pekalongan presents several challenges, including inefficiencies in attendance
recording, delays in administrative processes, and risks of errors and data
manipulation. Therefore, this study aims to develop a mobile-based Human Resource Information System
(HRIS) integrated with facial recognition technology using the Convolutional Neural Network (CNN)
method.
The development method used in this study follows the Waterfall model, which includes
requirements, design, development, testing, and deployment. The application is
built using the Flutter framework. The attendance system utilizes face detection and face verification technology based on
Google ML Kit and TensorFlow Lite to ensure accurate facial verification.
Testing results using the black
box testing method show that all application features function properly
according to the designed specifications. Based on the facial attendance system
testing using CNN with Cosine
Similarity and confusion matrix,
a 70% threshold achieved the highest accuracy of 0.77, indicating a good
balance between False Positive
and False Negative. Usability
evaluation using the System Usability
Scale (SUS) produced a score with an acceptable grade, indicating a high level of user acceptance.
The implementation of this system is expected to improve HRM efficiency at BMT
Bahtera Pekalongan, reduce the risk of attendance recording errors, and enhance
data transparency and reliability.
Keywords:
HRIS, CNN, Face Detection, Face Verification, Flutter, System Usability Scale,
Mobile Application