Browsing by Autor "Fathorazi Nur Fajri"

Now showing 1 - 3 of 3

Detection of Eight Skin Diseases Using Convolutional Neural Network with MobileNetV2 Architecture for Identification and Treatment Recommendation on Android Application
(2024) Ainul Furqon; Kamil Malík; Fathorazi Nur Fajri
Skin diseases are common in Indonesia due to the tropical climate, high population density, and low public awareness about skin health. These diseases are often caused by infections, chemical contamination, or other external factors and typically develop internally before becoming visible, with contact dermatitis being the most frequently reported condition. To address this issue, this research proposes the use of Artificial Intelligence (AI), specifically Convolutional Neural Network (CNN) with the MobileNetV2 architecture, to detect eight types of skin diseases, namely cellulitis, impetigo, athlete's foot, nail fungus, ringworm, cutaneous larva migrans, chickenpox, and shingles. MobileNetV2 was chosen for its efficiency and high accuracy in mobile applications. The methodology involves developing a detection system using CNN MobileNetV2, integrated into an Android application to identify skin diseases and provide treatment recommendations. The dataset was collected, labeled, resized, and normalized to meet the model requirements. After training, the model was tested using a separate dataset to ensure its generalization ability and was finally integrated into the Android application. This application allows users to detect skin diseases and receive treatment advice directly. The research results show that the CNN MobileNetV2 model achieves high accuracy in classifying the eight types of skin diseases, with stable performance over several training epochs. Evaluation of the test dataset revealed an overall accuracy of 97%, with high precision, recall, and F1-score for all disease classes. The application achieved an accuracy of 84% on general data, demonstrating its practical utility. However, the need for real-time updates of treatment information was identified as a limitation. This research advances skin disease detection technology and improves public access to accurate healthcare services. Future studies should focus on real-time treatment information updates and expanding the range of detectable diseases to enhance skin disease application.
Enhancing Low-Light PPE Violation Detection in Industrial Environments Using Multi-Contrast Image Processing and YOLOv9
(Engineering, Technology & Applied Science Research, 2025) Fathorazi Nur Fajri; Kamil Malik; Abu Tholib
Ensuring compliance with Personal Protective Equipment (PPE) is critical for workplace safety, yet low-light conditions reduce the accuracy of vision-based monitoring systems. This study introduces a novel violation-aware PPE dataset consisting of 11,407 annotated images across eight categories, explicitly modeling both compliance and non-compliance classes (e.g., helmet/no-helmet, vest/no-vest, gloves/no-gloves, shoes/no-shoes). The dataset differs from prior works by focusing on low-light industrial environments where detection is most challenging. To address these conditions, three multi-contrast enhancement techniques—Contrast Limited Adaptive Histogram Equalization (CLAHE), Auto-CLAHE, and EnlightenGAN—were integrated with an optimized YOLOv9t model. The six You Only Look Once (YOLO) variants (YOLOv5s–YOLOv12n) were benchmarked, with YOLOv9t plus augmentation achieving the best performance, with mAP@50 of 0.915, mAP@50–95 of 0.659, precision of 0.906, recall of 0.844, and inference time of 4.0 ms. The enhancement experiments demonstrated that CLAHE provided the highest detection coverage (79.43%), Auto-CLAHE yielded the greatest detection density (2.29/frame), and EnlightenGAN offered limited benefits due to domain shift. The findings confirm that histogram-based methods consistently improve PPE violation detection under low-light conditions, while Generative Adversarial Network (GAN)-based approaches require domain-specific adaptation. Overall, this study contributes a new dataset, systematic YOLO benchmarking, and the first task-driven comparison of classical, adaptive, and GAN-based enhancement methods for reliable, real-time workplace safety in challenging lighting environments.
Implementation of Personal Protective Equipment Detection Using Django and Yolo Web at Paiton Steam Power Plant (PLTU)
(2023) Khoirun Nisa; Fathorazi Nur Fajri; Zainal Arifin
Work accidents can occur at any time and unexpectedly, so work safety is associated with health because the work safety system in Indonesia is related to the K3 (Occupational Safety and Health) program. To create a safe and healthy work environment, occupational safety and health management are implemented to avoid work accidents by requiring every worker to use Personal Protective Equipment (PPE). This research aims to develop an immediate detection system for violations of Personal Protective Equipment (PPE) in the workplace using the Yolov8 Method and the Django web-based user interface framework. Yolov8 is one of the latest deep-learning object identification models while Django is the most popular Python developer framework. The system is designed to improve workplace safety and prevent accidents by monitoring compliance with PPE requirements. The research methodology involves literature study, image data collection, preprocessing, model training, and system deployment using the Django framework. There are four classes of detection based on the bounding box according to the specified color, the use of helmets and safety vests based on the red bounding box for helmets and blue for vests while when helmets and safety vests are not being used, based on green and yellow bounding boxes. The system successfully detected four PPE classes with an average accuracy of 82.3% from 230 test data, a mAP50 value of 81.6%, a precision value of 90.3%, and a recall value of 75.1%. The findings from this study indicate that the developed system can effectively improve occupational safety and health management. However, there is a detection error factor caused by the lighting and specifications of the camera used. Future research can focus on integrating the system with other work safety systems to provide a comprehensive solution for accident prevention.