Comparison of MobileNetV2 and VGG19 for the Categorization of Thermal Images

Object detection in thermal images is vital for diverse applications, utilizing machine learning algorithms to analyze infrared radiation. This study focuses on classifying contrast-enhanced thermal images (CLAHE) using MobileNetV2 and VGG19 architectures. Our research compares their performance metrics, emphasizing accuracy, F1 score, and recall. Results indicate MobileNetV2 outperforms VGG19, achieving 95% accuracy and 96.21% F1 score, whereas VGG19 scored 90.33% and 90.57%, respectively. This study demonstrates the effectiveness of MobileNetV2 in thermal image classification, showcasing its potential for real-world applications. Our findings underscore the applicability of transfer learning techniques in thermal image analysis, providing valuable insights for future research and development.