ANALISIS PENGARUH METODE IMAGE ENHANCEMENT KIND++ TERHADAP MODEL DETEKSI DAN PROFILING WAJAH PADA KONDISI LOW-LIGHT

Dewi Anggita Yulianti
Yana Gerhana Aditia
Gitarja Sandi


DOI: https://doi.org/10.29100/jipi.v10i3.8283

Abstract


Model deteksi dan profiling wajah menjadi salah satu teknologi pen-dukung dalam penanganan kasus kejahatan yang terus meningkat. Per-forma model deteksi dan profiling (estimasi usia, dan klasifikasi jenis kelamin) yang menurun secara signifikan dalam kondisi low-light, menjadi tantangan yang serius. Untuk mengatasi tantangan tersebut, penelitian ini menerapkan metode image enhancement berbasis deep learning, yaitu KinD++, sebagai tahap preprocessing sebelum gambar dianalisis oleh model. Proses enhancement berpotensi meningkatkan kualitas visual, namun juga memungkinkan terjadinya perubahan struktur wajah yang penting bagi proses analisis. Penelitian ini men-gevaluasi pengaruh KinD++ terhadap performa ketiga model dalam kondisi low-light. Berdasarkan hasil penelitian, penerapan metode en-hancement KinD++ terbukti memberikan dampak positif terhadap se-luruh model yang diuji. Pada model deteksi wajah, nilai mAP@50-95 meningkat dari 0,24 menjadi 0,28. Pada model klasifikasi jenis ke-lamin, akurasi meningkat dari 0,80 menjadi 0,82. Sementara itu, MAE (Mean Average Error) model estimasi usia mengalami penurunan dari 15,2 menjadi 7,6. Hasil ini menunjukkan bahwa KinD++ membantu meningkatkan performa model yang menurun akibat kondisi low-light, walaupun perubahan visual yang ditimbulkan tetap berisiko meng-ganggu informasi penting pada wajah, sehingga hasilnya tidak sebaik performa model saat pencahayaan normal.

Keywords


Deteksi Wajah; Peningkatan Citra; Pencaha-yaan Rendah; Profiling Wajah

Full Text:

PDF

Article Metrics :

References


Global Initiative Against Transnational Organized Crime. (2023). Global Organized Crime Index 2023. Global Initiative. Geneva, Switzerland. Diakses 13 April 2025. [Online]. Tersedia: https://globalinitiative.net/analysis/ocindex-2023/

Wang, X., Peng, J., Zhang, S., Chen, B., Wang, Y., & Guo, Y. (Desember 2022). A Survey of Face Recognition. arXiv preprint. [Online]. Tersedia: https://doi.org/10.48550/arXiv.2212.13038

Hangaragi, S., Singh, T., & Neelima, N. (Januari 2023). Face detection and recognition using face mesh and deep neural network. Procedia Computer Science, 218, 741–749. [Online]. Tersedia: https://doi.org/10.1016/j.procs.2023.01.054

Shen, W., Guo, Y., Wang, Y., Zhao, K., Wang, B., & Yuille, A. (Juni 2018). Deep Regression Forests for Age Estimation. Dalam Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2304–2313. [Online]. Tersedia: https://doi.org/10.1109/CVPR.2018.00245

Brown, E., & Perrett, D. I. (Juli 1993). What gives a face its gender? Perception, 22(7), 829–840. [Online]. Tersedia: https://doi.org/10.1068/p220829

Wang, W., Yang, W., & Liu, J. (2021). HLA-FACE: Joint high-low adaptation for low light face detection. Dalam Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 16190–16199. [Online]. Tersedia: https://doi.org/10.1109/CVPR46437.2021.01593

Rani, S., Pournima, S., Aram, A., Shanmuganeethi, V., Thiruselvan, P., & Rufus, N. (April 2024). Enhancing Facial Recognition Accuracy in Low-Light Conditions Using Convolutional Neural Networks. Journal of Electrical Systems, 20(5s), 2140–2148. [Online]. Tersedia: https://doi.org/10.52783/jes.2559

Pusiknas. (September 2023). Ratusan Kejahatan Ditindak dalam Waktu 10 Jam. Pusiknas Bareskrim Polri. Diakses 13 April 2025. [Online]. Tersedia: https://pusiknas.polri.go.id/detail_artikel/ratusan_kejahatan_ditindak_dalam_waktu_10_jam

Qiu, Y., Shi, Y., Lin, Y., Hu, Y., & Zhou, Z. (Januari 2024). A Survey on Image Enhancement and Restoration Techniques. Dalam Proceedings of the 3rd International Conference on Internet Technology and Educational Informatization (ITEI 2023), 24–26. [Online]. Tersedia: https://doi.org/10.4108/eai.24-11-2023.2343558

Paul, K. C., & Aslan, S. (Januari 2021). An Improved Real-Time Face Recognition System at Low Resolution Based on Local Binary Pattern Histogram Algorithm. Optics and Photonics, 11(4), 63–78. [Online]. Tersedia: https://doi.org/10.4236/opj.2021.114005

Guo, C., et al. (Januari 2020). Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement. arXiv preprint. [Online]. Tersedia: https://doi.org/10.48550/arXiv.2001.06826

Li, P. (Juli 2024). Multi-model fusion of DSFD and Zero-DCE for low-light face detection. Applied and Computational Engineering, 86(0), 108–115. [Online]. Tersedia: https://doi.org/10.54254/2755-2721/86/20241584

Zhang, Y., Guo, X., Ma, J., Liu, W., & Zhang, J. (April 2021). Beyond Brightening Low-light Images. International Journal of Computer Vision, 129(4), 1013–1037. [Online]. Tersedia: https://doi.org/10.1007/s11263-020-01407-x

Yu, J., Hao, X., & He, P. (2021). Single-stage Face Detection under Extremely Low-light Conditions. Dalam IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), 3516–3525. [Online]. Tersedia: https://doi.org/10.1109/ICCVW54120.2021.00392

Liang, J., et al. (Maret 2021). Recurrent Exposure Generation for Low-Light Face Detection. IEEE Transactions on Multimedia, 24, 1609–1621. [Online]. Tersedia: https://doi.org/10.1109/TMM.2021.3068840

Terven, J., Córdova-Esparza, D. M., & Romero-González, J. A. (November 2023). A Comprehensive Review of YOLO Architectures in Computer Vision: From YOLOv1 to YOLOv8 and YOLO-NAS. Machine Learning and Knowledge Extraction, 5(4), 1680–1716. [Online]. Tersedia: https://doi.org/10.3390/make5040083

Simonyan, K., & Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint. [Online]. Tersedia: https://doi.org/10.48550/arXiv.1409.1556

Rajeev, P. S., & Vishwakarma, G. (April 2024). Detection of Gender in Crowds Using ResNet Model. Journal of Electrical Systems, 20(2s), 389–408. [Online]. Tersedia: https://doi.org/10.52783/jes.1151

Yang, S., Luo, P., Loy, C. C., & Tang, X. (2016). WIDER FACE: A Face Detection Benchmark. Dalam IEEE Conference on Computer Vision and Pattern Recognition (CVPR). [Online]. Tersedia: https://doi.org/10.1109/CVPR.2016.596

Zhang, S. Y., & Qi, H. (Februari 2017). Age Progression/Regression by Conditional Adversarial Autoencoder. Dalam IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 5810–5818. [Online]. Tersedia: https://doi.org/10.48550/arXiv.1702.08423