Attendance System based on Face Recognition, Face Mask and Body Temperature Detection on Raspberry Pi

The world is currently experiencing a COVID-19 pandemic which has caused many deaths. Then, one of COVID-19 early detection can be done through wearing a face mask and detecting body temperature when entering a room. The purpose of this research is to create an attendance system capable to recognize face, face mask, and measure body temperature. The data was tested against three types of Raspberry Pi (Pi 3B, Pi 4-4Gb, and Pi 4-8Gb). Each type of Raspberry Pi was tested using three metric combinations of different encoding methods and object classification, namely Method 1 - Haar Cascade and LBPH, Method 2 - Haar Cascade, and Tensorflow, Method 3 - MTCNN and Tensorflow. The test results obtained Method 1: FPS 15, accuracy rate 60%, CPU temperature 58°C, Method 2: FPS 4, accuracy rate 90%, CPU temperature 65°C, Method 3: FPS 2, accuracy rate 95%, CPU temperature 68°C. Based on the benchmarks and scoring, it can be concluded that the most optimal attendance system based on face recognition, mask, and body temperature detection is a system with a Raspberry Pi 4-4Gb and Method 1-Haar Cascade and LBPH. For next development, this attendance system needs to be further improved in security and accuracy without increasing CPU load.