Multi-Seed Robustness Benchmark of Lightweight YOLO Models for Young Crescent Moon Detection under Limited-Data Conditions
DOI:
https://doi.org/10.63158/journalisi.v8i3.1653Keywords:
young crescent moon detection, lightweight object detection, YOLO benchmark, multi-seed reproducibility, CPU inferenceAbstract
Visual observation of the young crescent moon is challenging due to its thin and low-contrast appearance. Although YOLO-based object detectors are promising for image-based crescent localization, newer architectures do not automatically generalize well to small grayscale datasets, and prior studies rarely report robustness across repeated training runs. This study benchmarks YOLOv8n, YOLO11n, and YOLO26n for young crescent moon detection under limited-data conditions. A grayscale dataset of 697 images was resized to 640 × 640 pixels, annotated with the single class crescent_moon, and split into training, validation, and test subsets at a fixed 70:20:10 ratio. The three models were trained using the same configuration across five random seeds. Validation results were used to analyze multi-seed robustness, while the fixed 71-image test set was used for CPU-only inference evaluation. YOLO26n achieved the highest validation mAP@50-95 and fitness with the lowest variability, and also achieved the lowest CPU pipeline latency and highest throughput on the test set. These findings show that YOLO26n offers the best trade-off between accuracy and efficiency across the evaluated dataset and CPU-only inference setting. The reported throughput reflects low-frame-rate image-based inference, not real-time video performance. This study provides a reproducible benchmark protocol that combines fixed data splitting, grayscale preprocessing, data integrity checking, multi-seed robustness analysis, and CPU inference profiling.
Downloads
References
[1] W. N. J. Hj Wan Yussof, M. Man, R. Umar, A. N. Zulkeflee, E. A. Awalludin, and N. Ahmad, "Enhancing Moon Crescent Visibility Using Contrast-Limited Adaptive Histogram Equalization and Bilateral Filtering Techniques," Journal of Telecommunications and Information Technology, vol. 2022, no. 1, pp. 3–13, Mar. 2022, doi: 10.26636/JTIT.2022.155721.
[2] R. Muztaba, H. L. Malasan, and M. Djamal, "Deep learning for crescent detection and recognition: Implementation of Mask R-CNN to the observational Lunar dataset collected with the Robotic Lunar Telescope System," Astronomy and Computing, vol. 45, p. 100757, Oct. 2023, doi: 10.1016/J.ASCOM.2023.100757.
[3] Maskufa, Sopa, S. Hidayati, and A. Damanhuri, "Implementation of the New MABIMS Crescent Visibility Criteria: Efforts to Unite the Hijriyah Calendar in the Southeast Asian Region," AHKAM : Jurnal Ilmu Syariah, vol. 22, no. 1, pp. 209–236, Jun. 2022, doi: 10.15408/AJIS.V22I1.22275.
[4] A. Mufid and T. Djamaluddin, "The implementation of new minister of religion of Brunei, Indonesia, Malaysia, and Singapore criteria towards the Hijri calendar unification," HTS Teologiese Studies / Theological Studies, vol. 79, no. 1, p. 8, Jun. 2023, doi: 10.4102/HTS.V79I1.8774.
[5] J. Terven, D.-M. Córdova-Esparza, and J.-A. Romero-González, "A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NAS," Mach. Learn. Knowl. Extr., vol. 5, no. 4, pp. 1680–1716, Nov. 2023, doi: 10.3390/make5040083.
[6] Z. Wang et al., "PC-YOLO11s: A Lightweight and Effective Feature Extraction Method for Small Target Image Detection," Sensors, vol. 25, no. 2, p. 348, Jan. 2025, doi: 10.3390/s25020348.
[7] W. D. Jima, S. F. Desta, T. A. Tarekegn, G. S. Gebremedhin, A. B. Gutema, and T. G. Debelee, "Tsetse Fly Detection and Sex Classification Model Enrichment Employing YOLOv8 and YOLO11 Architecture," Applied AI Letters, vol. 6, no. 3, p. e70004, Oct. 2025, doi: 10.1002/ail2.70004.
[8] H.-V. Nguyen, J.-H. Bae, Y.-E. Lee, H.-S. Lee, and K.-R. Kwon, "Comparison of Pre-Trained YOLO Models on Steel Surface Defects Detector Based on Transfer Learning with GPU-Based Embedded Devices," Sensors, vol. 22, no. 24, p. 9926, Dec. 2022, doi: 10.3390/s22249926.
[9] D. Dai, J. Gao, S. Parsons, and E. Sklar, "Small datasets for fruit detection with transfer learning," UKRAS21 Conference: Robotics at home Proceedings, vol. 4, pp. 5–6, Jul. 2021, doi: 10.31256/NF6UH8Q.
[10] M. Al-Rajab, S. Loucif, and Y. Al Risheh, "Predicting new crescent moon visibility applying machine learning algorithms," Sci. Rep., vol. 13, no. 1, p. 6674, Apr. 2023, doi: 10.1038/s41598-023-32807-x.
[11] S. Loucif, M. Al-Rajab, R. Abu Zitar, and M. Rezk, "Toward a globally lunar calendar: a machine learning-driven approach for crescent moon visibility prediction," J. Big Data, vol. 11, no. 1, p. 114, Aug. 2024, doi: 10.1186/s40537-024-00979-6.
[12] Z. T. Allawi, "Crescent Moon Visibility: A New Criterion using Deep learned Artificial Neural-Network," Iraqi Journal of Science, pp. 2332–2343, Apr. 2024, doi: 10.24996/ijs.2024.65.4.45.
[13] A. L. A. Mohd Nasir et al., "Comparative Analysis of Image Processing Technique in Determining the New Crescent Moon Visibility," J. Phys. Conf. Ser., vol. 2915, no. 1, p. 012004, Dec. 2024, doi: 10.1088/1742-6596/2915/1/012004.
[14] R. Muztaba, H. L. Malasan, and M. Djamal, "A Self-Construction of Automatic Crescent Detection Using Haar-Cascade Classifier and Support Vector Machine," J. Phys. Conf. Ser., vol. 2734, no. 1, p. 012007, Mar. 2024, doi: 10.1088/1742-6596/2734/1/012007.
[15] J. A. Utama et al., "Young lunar crescent detection based on video data with computer vision techniques," Astronomy and Computing, vol. 44, p. 100731, Jul. 2023, doi: 10.1016/J.ASCOM.2023.100731.
[16] D. Picard, "Torch.manual_seed(3407) is all you need: On the influence of random seeds in deep learning architectures for computer vision," arXiv:2109.08203, May 2023. Accessed: May 06, 2026. [Online]. Available: http://arxiv.org/abs/2109.08203
[17] M. Rosenblatt, L. Tejavibulya, R. Jiang, S. Noble, and D. Scheinost, "Data leakage inflates prediction performance in connectome-based machine learning models," Nat. Commun., vol. 15, no. 1, p. 1829, Feb. 2024, doi: 10.1038/s41467-024-46150-w.
[18] E. P. Silmina, S. Sunardi, and A. Yudhana, "COMPARATIVE ANALYSIS OF YOLO DEEP LEARNING MODEL FOR IMAGE-BASED BEEF FRESHNESS DETECTION," JITK (Jurnal Ilmu Pengetahuan dan Teknologi Komputer), vol. 11, no. 1, pp. 250–265, Aug. 2025, doi: 10.33480/JITK.V11I1.6784.
[19] Pexels, "Young Crescent Search Results," 2026. Accessed: May 24, 2026. [Online]. Available: https://www.pexels.com/search/young%20crescent/
[20] Badan Meteorologi Klimatologi dan Geofisika, “Galeri Pengamatan Hilal,” 2026. Accessed: May 24, 2026. [Online]. Available: https://hilal.bmkg.go.id/gallery
[21] M. Elsässer, “Mondbeobachtungen / Observations of the Moon,” 2025. Accessed: May 24, 2026. [Online]. Available: https://www.mondatlas.de/beob_sicheln.html
[22] International Astronomy Center, "Islamic Crescents' Observation Project (ICOP)," 2026. Accessed: May 24, 2026. [Online]. Available: https://www.icoproject.org/
[23] Pexels, "Free Stock Photo and Video License," 2026. Accessed: May 24, 2026. [Online]. Available: https://www.pexels.com/license/
[24] P. Skalski, “Makesense.ai,” 2019. Accessed: May 24, 2026. [Online]. Available: https://www.makesense.ai/
[25] C. Wang, Q. Wang, Y. Qian, Y. Hu, Y. Xue, and H. Wang, "DP-YOLO: Effective Improvement Based on YOLO Detector," Applied Sciences, vol. 13, no. 21, p. 11676, Oct. 2023, doi: 10.3390/app132111676.
[26] Q. Xu, Y. Li, and Z. Shi, "LMO-YOLO: A Ship Detection Model for Low-Resolution Optical Satellite Imagery," IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 15, pp. 4117–4131, 2022, doi: 10.1109/JSTARS.2022.3176141.
[27] X. Li, L. Chang, and X. Liu, "CE-Dedup: Cost-Effective Convolutional Neural Nets Training based on Image Deduplication," in Proc. IEEE ISPA/BDCloud/SocialCom/SustainCom, IEEE, Sep. 2021, pp. 11–18. doi:
10.1109/ISPA-BDCloud-SocialCom-SustainCom52081.2021.00017.
[28] R. Sapkota, R. H. Cheppally, A. Sharda, and M. Karkee, "YOLO26: Key Architectural Enhancements and Performance Benchmarking for Real-Time Object Detection," arXiv:2509.25164, Mar. 2026. Accessed: May 06, 2026. [Online]. Available: http://arxiv.org/abs/2509.25164
[29] P. Hidayatullah and R. Tubagus, "YOLO26: A Comprehensive Architecture Overview and Key Improvements," arXiv:2602.14582, Feb. 2026. Accessed: May 06, 2026. [Online]. Available: http://arxiv.org/abs/2602.14582
[30] P. Hidayatullah, N. Syakrani, M. R. Sholahuddin, T. Gelar, and R. Tubagus, “YOLOv8 to YOLO11: A Comprehensive Architecture In-depth Comparative Review,” Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi), vol. 10, no. 2, pp. 341–354, Apr. 2025, doi: 10.29207/RESTI.V10I2.6598.
[31] Ultralytics, "Ultralytics YOLO Documentation," 2026. Accessed: May 24, 2026. [Online]. Available: https://docs.ultralytics.com/
[32] P. Ruiz-Ponce, D. Ortiz-Perez, J. Garcia-Rodriguez, and B. Kiefer, "POSEIDON: A Data Augmentation Tool for Small Object Detection Datasets in Maritime Environments," Sensors, vol. 23, no. 7, p. 3691, Apr. 2023, doi: 10.3390/s23073691.
[33] L. Liao, L. Luo, J. Su, Z. Xiao, F. Zou, and Y. Lin, "Eagle-YOLO: An Eagle-Inspired YOLO for Object Detection in Unmanned Aerial Vehicles Scenarios," Mathematics, vol. 11, no. 9, p. 2093, Apr. 2023, doi: 10.3390/math11092093.
[34] J.-H. Kim, N. Kim, Y. W. Park, and C. S. Won, "Object Detection and Classification Based on YOLO-V5 with Improved Maritime Dataset," J. Mar. Sci. Eng., vol. 10, no. 3, p. 377, Mar. 2022, doi: 10.3390/jmse10030377.
[35] E. Thibeau-Sutre et al., "ClinicaDL: An open-source deep learning software for reproducible neuroimaging processing," Comput. Methods Programs Biomed., vol. 220, p. 106818, Jun. 2022, doi: 10.1016/J.CMPB.2022.106818.
[36] A. Ghezal and A. König, "A Comparative Study of Hybrid Machine-Learning vs. Deep-Learning Approaches for Varroa Mite Detection and Counting," Sensors, vol. 25, no. 16, p. 5075, Aug. 2025, doi: 10.3390/s25165075.
[37] D. Kolosov, V. Kelefouras, P. Kourtessis, and I. Mporas, "Anatomy of Deep Learning Image Classification and Object Detection on Commercial Edge Devices: A Case Study on Face Mask Detection," IEEE Access, vol. 10, pp. 109167–109186, 2022, doi: 10.1109/ACCESS.2022.3214214.
[38] L. Wood and F. Chollet, "Efficient Graph-Friendly COCO Metric Computation for Train-Time Model Evaluation," arXiv:2207.12120, Jul. 2022. Accessed: May 06, 2026. [Online]. Available: http://arxiv.org/abs/2207.12120
[39] D. Einsiedel, M. Vita, F. Kaltenecker, B. Dunnewind, J. Meulendijks, and C. Krupitzer, "Automated Detection of Quality Deviations in Poultry Processing Using Step-Specific YOLOv12 Models," Foods, vol. 15, no. 6, p. 1019, Mar. 2026, doi: 10.3390/foods15061019.
[40] Ultralytics, "Reference for ultralytics/utils/metrics.py - Ultralytics YOLO Docs." Accessed: May 06, 2026. [Online]. Available: https://docs.ultralytics.com/reference/utils/metrics/#ultralytics.utils.metrics.ClassifyMetrics.fitness
[41] M. Zhang, S. Xu, W. Song, Q. He, and Q. Wei, "Lightweight Underwater Object Detection Based on YOLO v4 and Multi-Scale Attentional Feature Fusion," Remote Sens. (Basel)., vol. 13, no. 22, p. 4706, Nov. 2021, doi: 10.3390/rs13224706.
[42] Z. Huang, J. Wu, L. Su, Y. Xie, T. Li, and X. Huang, "SP-YOLO-Lite: A Lightweight Violation Detection Algorithm Based on SP Attention Mechanism," Electronics (Basel)., vol. 12, no. 14, p. 3176, Jul. 2023, doi: 10.3390/electronics12143176.
[43] J. Cao, W. Bao, H. Shang, M. Yuan, and Q. Cheng, "GCL-YOLO: A GhostConv-Based Lightweight YOLO Network for UAV Small Object Detection," Remote Sens. (Basel)., vol. 15, no. 20, p. 4932, Oct. 2023, doi: 10.3390/rs15204932.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Journal of Information Systems and Informatics
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors Declaration
- The Authors certify that they have read, understood, and agreed to the Journal of Information Systems and Informatics (JournalISI) submission guidelines, policies, and submission declaration. The submission has been prepared using the provided template.
- The Authors certify that all authors have approved the publication of this manuscript and that there is no conflict of interest.
- The Authors confirm that the manuscript is their original work, has not received prior publication, is not under consideration for publication elsewhere, and has not been previously published.
- The Authors confirm that all authors listed on the title page have contributed significantly to the work, have read the manuscript, attest to the validity and legitimacy of the data and its interpretation, and agree to its submission.
- The Authors confirm that the manuscript is not copied from or plagiarized from any other published work.
- The Authors declare that the manuscript will not be submitted for publication in any other journal or magazine until a decision is made by the journal editors.
- If the manuscript is finally accepted for publication, the Authors confirm that they will either proceed with publication immediately or withdraw the manuscript in accordance with the journal’s withdrawal policies.
- The Authors agree that, upon publication of the manuscript in this journal, they transfer copyright or assign exclusive rights to the publisher, including commercial rights
