Analisis Peningkatan Kinerja Model InceptionResNetV2 dalam Identifikasi Wajah pada Dataset In-the-Wild melalui Optimasi Hyperparameter

  • Ferry Febiansah Nusa Putra University ORCID
PDF
DOI https://doi.org/10.31544/jtera.v11.i1.2026.69-76
Published 2026-06-26
Full Text Preview
Download PDF

Abstract

Deep learning-powered facial recognition is essential across various applications, including security systems, authentication processes, and identity verification. However, model performance on in-the-wild datasets faces challenges due to variations in lighting conditions, pose differences, diverse facial expressions, and background complexity. This study aims to analyze the performance improvement of the InceptionResNetV2 model through hyperparameter optimization for the face identification task, using a primary dataset comprising 5,180 face images of 153 individuals captured in in‑the‑wild conditions. The model employs transfer learning, utilizing initial weights from ImageNet, with fine-tuning applied to the final 50 layers. Hyperparameter optimization uses the grid search method, combining learning rates of 0.001 and 0.0005, and batch sizes of 8 and 16. Model performance is evaluated using accuracy, precision, recall, and F1-score metrics, supported by confusion matrix analysis and prediction case visualizations. Results show that the optimized model improved accuracy from 0.8830 to 0.9148, with consistent gains in precision (0.9063 to 0.9249), recall (0.9016 to 0.9302), and F1-score (0.8839 to 0.9139). Confusion matrix analysis indicates an increase in correct predictions across different classes, while improvement case visualizations demonstrate the model's ability to correct prediction errors under complex conditions. McNemar's statistical test yields a p-value of 0.000639 (< 0.05), indicating that the performance improvement is statistically significant. Therefore, hyperparameter optimization is effective in enhancing model performance on in-the-wild datasets.

 

Keywords

identifikasi wajah deep learning InceptionResNetV2 hyperparameter tuning transfer learning

References

S. Khan, M. Naseer, M. Hayat, S. W. Zamir, F. S. Khan, and M. Shah, “Transformers in Vision: A Survey,” ACM Comput. Surv., vol. 54, no. 10s, pp. 1–41, Jan. 2022, doi: 10.1145/3505244.
V. Tomar, N. Kumar, and A. R. Srivastava, “Single sample face recognition using deep learning: a survey,” Artif. Intell. Rev., vol. 56, no. S1, pp. 1063–1111, Oct. 2023, doi: 10.1007/s10462-023-10551-y.
M. Wang and W. Deng, “Deep face recognition: A survey,” Neurocomputing, vol. 429, pp. 215–244, Mar. 2021, doi: 10.1016/j.neucom.2020.10.081.
K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image Recognition,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, Jun. 2016, pp. 770–778. doi: 10.1109/CVPR.2016.90.
T. Yu and H. Zhu, “Hyper-Parameter Optimization: A Review of Algorithms and Applications,” 2020. doi: 10.48550/arxiv.2003.05689.
C. Zhang, S. Bengio, M. Hardt, B. Recht, and O. Vinyals, “Understanding deep learning requires rethinking generalization,” 2017. [Online]. Available: https://arxiv.org/abs/1611.03530
M. Feurer and F. Hutter, “Hyperparameter Optimization,” 2019, pp. 3–33. doi: 10.1007/978-3-030-05318-5_1.
D. Chicco and G. Jurman, “The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation,” BMC Genomics, vol. 21, no. 1, p. 6, Dec. 2020, doi: 10.1186/s12864-019-6413-7.
G. Sari, S. B. Okcu, D. Demir, İ. Göçer, and C. Çiğla, “Face recognition in the wild: augmentation-based robustness against real-world degradations,” in Artificial Intelligence for Security and Defence Applications III, Y. Yitzhaky, R. Prabhu, and H. J. Kuijf, Eds., SPIE, Oct. 2025, p. 20. doi: 10.1117/12.3069782.
Y. Huang, Z. Wu, J. Chen, and H. Xiang, “Privacy-Preserving Face Recognition Method Based on Randomization and Local Feature Learning,” J. Imaging, vol. 10, no. 3, p. 59, Feb. 2024, doi: 10.3390/jimaging10030059.
Diniati Ruaika and Shofwatul Uyun, “Optimisation of Residual Network Using Data Augmentation and Ensemble Deep Learning for Butterfly Image Classification,” IJID (International J. Informatics Dev., vol. 12, no. 2, pp. 350–361, Jan. 2024, doi: 10.14421/ijid.2023.4038.
Y. Jaber, P. Dharmasena, A. Nassif, and N. Nassif, “Hyperparameter Optimization of Neural Networks Using Grid Search for Predicting HVAC Heating Coil Performance,” Buildings, vol. 15, no. 15, p. 2753, Aug. 2025, doi: 10.3390/buildings15152753.
N. H. Tiep et al., “A New Hyperparameter Tuning Framework for Regression Tasks in Deep Neural Network: Combined-Sampling Algorithm to Search the Optimized Hyperparameters,” Mathematics, vol. 12, no. 24, p. 3892, Dec. 2024, doi: 10.3390/math12243892.
L. Xu, X. Chen, and X. Yang, “Tourism image classification based on convolutional neural network SqueezeNet——Taking Slender West Lake as an example,” PLoS One, vol. 19, no. 1, p. e0295439, Jan. 2024, doi: 10.1371/journal.pone.0295439.
O. Rainio, J. Teuho, and R. Klén, “Evaluation metrics and statistical tests for machine learning,” Sci. Rep., vol. 14, no. 1, p. 6086, Mar. 2024, doi: 10.1038/s41598-024-56706-x.
K. L. Tan, C. P. Lee, and K. M. Lim, “A Survey of Sentiment Analysis: Approaches, Datasets, and Future Research,” Appl. Sci., vol. 13, no. 7, p. 4550, Apr. 2023, doi: 10.3390/app13074550.

Issue

Vol. 11 No. 1: June 2026

Section

Articles

License

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Authors who publish with this journal agree to the following terms:

The copyright to this article is transferred to JTERA if and when the article is accepted for publication. The undersigned hereby transfers any and all rights in and to the paper including without limitation all copyrights to JTERA. The undersigned hereby represents and warrants that the paper is original and that he/she is the author of the paper, except for material that is clearly identified as to its original source, with permission notices from the copyright owners where required. The undersigned represents that he/she has the power and authority to make and execute this assignment. The copyright transfer form will be sent via email before the article published.

The corresponding author signs for and accepts responsibility for releasing this material on behalf of any and all co-authors. This agreement is to be signed by at least one of the authors who have obtained the assent of the co-author(s) where applicable. After submission of this agreement signed by the corresponding author, changes of authorship or in the order of the authors listed will not be accepted.

How to Cite

[1]
“Analisis Peningkatan Kinerja Model InceptionResNetV2 dalam Identifikasi Wajah pada Dataset In-the-Wild melalui Optimasi Hyperparameter”, JTERA, vol. 11, no. 1, pp. 69–76, Jun. 2026, doi: 10.31544/jtera.v11.i1.2026.69-76.