Claim Missing Document
Check
Articles

Implementasi Deep Learning Berbasis MobileNetV2 untuk Deteksi Real-Time Bacterial Spot dengan Pendekatan Arsitektur Lightweight Ahmad Nabilul As'ad; Elkaf Rahmawan Pramudya
Building of Informatics, Technology and Science (BITS) Vol 7 No 4 (2026): March 2026
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i4.9252

Abstract

Bacterial spot caused by Xanthomonas campestris pv. vesicatoria is a critical disease in bell peppers that can reduce productivity by up to 50%. This study implements MobileNetV2 with two-stage transfer learning for real-time bacterial spot detection using lightweight architecture approach, with ResNet50 as baseline comparison. PlantVillage dataset (2,475 images) was used for training and in-domain evaluation, while India dataset (132 images) for domain shift assessment. Results demonstrate MobileNetV2 achieves 98.66% accuracy on PlantVillage test set, outperforming ResNet50 (89.78%) by 8.88 percentage points despite being 9.2× lighter (2.7 MB vs 24.3 MB TFLite) and 2.0× faster (22.4 ms vs 45.8 ms inference time). MobileNetV2 efficiency advantage is also evident in its inference memory footprint of only 107 MB RAM, significantly 2.3x lower than ResNet50(242 MB RAM), making it highly suitable for deployment on mid-range smartphones with limited RAM. External dataset evaluation reveals MobileNetV2 maintains superior robustness with 65.3% retention rate versus ResNet50's 52.3%. Trade-off analysis positions MobileNetV2 on the Pareto frontier, achieving optimal accuracy-efficiency sweet spot for plant disease detection applications. This research contributes empirical evidence for lightweight architecture superiority, comprehensive efficiency-oriented evaluation framework, ULTRA-LIGHT training strategy for addressing inverse overfitting, and realistic generalization assessment using tropical external dataset.
Improved Lorenz chaotic map with dynamic zig-zag pattern for high-security digital image encryption Ajib Susanto; Elkaf Rahmawan Pramudya; Noorayisahbe Mohd Yaacob
Bulletin of Electrical Engineering and Informatics Vol 15, No 5: October 2026
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eei.v15i5.11639

Abstract

This study proposes a secure image encryption framework based on an improved Lorenz chaotic map combined with a dynamic zig-zag permutation strategy. The improved chaotic mapping introduces dynamic perturbations to increase the complexity of chaotic trajectories and expand the key space, while the dynamic zig-zag traversal enhances pixel-level diffusion and minimizes residual pixel correlations during the permutation stage. Experimental evaluation was conducted on multiple RGB images with a resolution of 512×512 pixels. The proposed method achieved mean squared error (MSE) values of approximately 1800–1900 and peak signal-to-noise ratio (PSNR) values around 15.8 dB, indicating significant structural distortion between plaintext and cipher images. In addition, the obtained unified average changing intensity (UACI) (~33.4%) and number of pixels change rate (NPCR) (~99.61%) values approach the theoretical ideals, demonstrating strong resistance against statistical and differential attacks. Comparative analysis shows that the proposed framework provides stronger statistical robustness than conventional Lorenz-based encryption schemes. Furthermore, the decryption process successfully reconstructs the original images without information loss, confirming the correctness and reversibility of the scheme. These results indicate that integrating dynamic perturbations with adaptive zig-zag permutation effectively improves diffusion strength, key space complexity, and overall encryption robustness for secure multimedia image transmission.
Enhancing Medical Image Security Using Hyperchaotic Lorenz and Josephus Traversing Encryption Eko Hari Rachmawanto; Elkaf Rahmawan Pramudya; Zudha Pratama
Scientific Journal of Informatics Vol. 11 No. 4: Special Issue
Publisher : Universitas Negeri Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15294/sji.v11i4.9815

Abstract

Purpose: The present work focuses on developing a methodology to encrypt medical images using combined Hyperchaotic Lorenz systems with Josephus Traversing. This, therefore, forms the basis of the present paper to establish the efficacy of the proposed method against glioma, meningioma, and pituitary kinds of brain tumor images at 256 × 256 and 512 × 512 pixels image sizes. Methods: In this regard, a state-of-the-art encryption technique based on the Hyperchaotic Lorenz systems for Josephus Traversing has been proposed against the medical images of glioma, meningioma, and pituitary tumor datasets obtained from the repository via medical imaging. Result: The different distortion of test outcomes has the MSE value lying between 69.01 and 172.1, while fidelity preservation-PSNR lies between 12.971 and 18.321 dB for different tumor types and sizes of images. The UACI is between 3.625 and 11.34, while the NPCR is always greater than 99% to show very high tamper resistance. This approach is very new in integrating chaos and traversal algorithms for encrypting medical images. Hence, it has a great promising enhancement of security and protection of patient privacy. Novelty: This research contributes a comprehensive investigation based on different metrics that allows exploring not only the efficiency but also strength against decryption techniques for a proposed encryption method. More investigations could be done for further research work in order to enhance the encryption speed, which would improve robustness against advanced decryption techniques in medical image security for digital health applications.