Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Jurnas Nasional Teknologi dan Sistem Informasi
Muhammad Reza Habibi
Department of Business Statistics, Faculty of Vocational Studies, Institut Teknologi Sepuluh Nopember (ITS)
Computer Science & IT Decision Sciences, Operations Research & Management

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Comparison of FA-SSD: Performance of Feature Extractors VGG-19, MobileNetV3, and ResNet-152 for Human Body Temperature Prediction Ridho Sholehurrohman; Akmal Junaidi; Hamzah Hanif; Favorisen Rosyking Lumbanraja; Muhammad Reza Habibi
Jurnal Nasional Teknologi dan Sistem Informasi Vol 12 No 2 (2026): Agustus 2026
Publisher : Departemen Sistem Informasi, Fakultas Teknologi Informasi, Universitas Andalas

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.25077/TEKNOSI.v12i2.2026.300-311

Abstract

Body temperature is a critical health indicator, and non-contact thermal imaging-based prediction systems are essential for early detection of infectious diseases. This study compares the performance of three FA-SSD (Feature Fusion and Spatial Attention-Based Single Shot Detector) models with different feature extractors—VGG-19, MobileNetV3, and ResNet-152—for face detection and body temperature prediction using thermal images. A subset of 520 thermal facial images from the Comprehensive Facial Thermal Dataset was used, with 80% for training, 10% for validation, and 10% for testing. Model performance was evaluated using Generalized Intersection over Union (GIoU) for detection accuracy, and Mean Absolute Error (MAE) with Mean Absolute Percentage Error (MAPE) for temperature prediction accuracy. The results showed that no single model excelled in all metrics. ResNet-152 achieved the highest average GIoU, indicating superior object detection performance. VGG-19 delivered the lowest average MAE of 0.451°C and MAPE of 1.278%, making it the best for temperature prediction. MobileNetV3 achieved the lowest minimum MAE of 0.000050°C but showed higher average errors and validation fluctuations. In conclusion, VGG-19 is recommended for clinical temperature accuracy, ResNet-152 for robust detection, and MobileNetV3 for edge deployment.
Co-Authors Akmal Junaidi Hamzah Hanif Lumbanraja, Favorisen R Ridho Sholehurrohman