Article Sidebar

Download
PDF

Main Article Content

Abstract

Bobot komponen kerugian kerap dianggap sebagai rincian pelatihan yang sepele, padahal pada sistem penghitungan kerumunan yang dibangun di atas tulang punggung difusi beku hanya kepala kepadatan yang benar-benar dilatih, sehingga perimbangan antara kerugian relatif, kehalusan lokal, dan penyelarasan grid menentukan mutu peta kepadatan yang dihasilkan. Penelitian ini memetakan konfigurasi supervisi yang dibutuhkan lima dataset tolok ukur dengan tingkat kepadatan dan keterandalan anotasi berbeda, lalu menghubungkan tiap konfigurasi dengan sifat anotasi titiknya. Adegan yang sangat padat menuntut regularisasi struktural terkuat dengan bobot kehalusan lokal dan penyelarasan grid sebesar nol koma sepuluh, sedangkan dataset beranotasi tidak rapi menuntut penonaktifan kerugian relatif dan penyelarasan grid disertai laju pembelajaran terkecil. Dengan konfigurasi tersebut galat absolut rata-rata yang dicapai adalah 51,39 dan 5,79 pada dua bagian ShanghaiTech, 75,19 pada UCF-QNRF, serta 56,21 pada JHU-Crowd++, sehingga tetap rapat terhadap metode penghitungan terkini pada tiga dataset pertama namun tertinggal pada dataset terakhir. Temuan tersebut menghasilkan pedoman penyetelan yang menempatkan bobot supervisi berbasis skala dan supervisi tingkat grid sebagai turunan dari keterandalan anotasi titik, bukan semata dari tingkat kepadatan adegan.

Keywords

bobot supervisi fungsi kerugian mutu anotasi penghitungan kerumunan

Article Details

License

Copyright (c) 2025 Mas Nurul Achmadiah, Muhammad Ridha Agam

Creative Commons License

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

References

  1. M. N. Achmadiah, N. Setyawan, A. A. Bryantono, C.-C. Sun, and W.-K. Kuo, “Fast Person Detection Using YOLOX With AI Accelerator For Train Station Safety,” in 2024 International Electronics Symposium (IES), IEEE, Aug. 2024, pp. 504–509. doi: 10.1109/IES63037.2024.10665874.
  2. Y. Zhang, D. Zhou, S. Chen, S. Gao, and Y. Ma, “Single-Image Crowd Counting via Multi-Column Convolutional Neural Network,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, Jun. 2016, pp. 589–597. doi: 10.1109/CVPR.2016.70.
  3. Y. Li, X. Zhang, and D. Chen, “CSRNet: Dilated Convolutional Neural Networks for Understanding the Highly Congested Scenes,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, IEEE, Jun. 2018, pp. 1091–1100. doi: 10.1109/CVPR.2018.00120.
  4. M. N. Achmadiah, C.-C. Sun, W.-K. Kuo, and J.-W. Hsieh, “RepSFNet : A Single Fusion Network with Structural Reparameterization for Crowd Counting,” in 2025 IEEE International Conference on Advanced Visual and Signal-Based Systems (AVSS), IEEE, Aug. 2025, pp. 1–6. doi: 10.1109/AVSS65446.2025.11149882.
  5. Boyu Wang, Huidong Liu, Dimitris Samaras, and Minh Hoai, “Distribution Matching for Crowd Counting,” in 34th Conference on Neural Information Processing Systems (NeurIPS 2020), Vancouver, Canada., 2020.
  6. Alec Radford * 1 Jong Wook Kim * 1 Chris Hallacy 1 Aditya Ramesh 1 Gabriel Goh 1 Sandhini Agarwal 1 Girish Sastry 1 Amanda Askell 1 Pamela Mishkin 1 Jack Clark 1 Gretchen Krueger 1 Ilya Sutskever 1, “Learning Transferable Visual Models From Natural Language Supervision,” in 38 th International Conference on Machine Learning, 2021.
  7. Y. Ma, V. Sanchez, and T. Guha, “CLIP-EBC: CLIP Can Count Accurately through Enhanced Blockwise Classification,” in 2025 IEEE International Conference on Multimedia and Expo (ICME), IEEE, Jun. 2025, pp. 1–6. doi: 10.1109/ICME59968.2025.11209839.
  8. N. Amini-Naieni, T. Han, and A. Zisserman, “CountGD: Multi-Modal Open-World Counting,” in Advances in Neural Information Processing Systems 37, San Diego, California, USA: Neural Information Processing Systems Foundation, Inc. (NeurIPS), 2024, pp. 48810–48837. doi: 10.52202/079017-1547.
  9. Yihong wu, jinqiau wei, xionghui zhao, and yidi li, “DSGC-Net: A Dual-Stream Graph Convolutional Network for Crowd Counting via Feature Correlation Mining,”
  10. N. Setyawan, C.-C. Sun, M.-H. Hsu, W.-K. Kuo, and J.-W. Hsieh, “MicroViT: A Vision Transformer with Low Complexity Self Attention for Edge Device,” in 2025 IEEE International Symposium on Circuits and Systems (ISCAS), IEEE, May 2025, pp. 1–5. doi: 10.1109/ISCAS56072.2025.11043206.
  11. M. Nurul Achmadiah, A. Ahamad, C.-C. Sun, and W.-K. Kuo, “Energy-Efficient Fast Object Detection on Edge Devices for IoT Systems,” IEEE Internet Things J., vol. 12, no. 11, pp. 16681–16694, Jun. 2025, doi: 10.1109/JIOT.2025.3536526.
  12. G. Al Azhar, S. Sungkono, M. N. Achmadiyah, and S. Izza, “Peningkatan Kestabilan Sistem Kontrol UGV melalui Optimalisasi Manajemen Core dan Free-RTOS pada ESP32,” Jurnal Elektronika dan Otomasi Industri, vol. 10, no. 2, pp. 253–263, Jul. 2023, doi: 10.33795/elkolind.v10i2.3720.
  13. N. Setyawan, M. N. Achmadiah, C.-C. Sun, and W.-K. Kuo, “Multi-Stage Vision Transformer for Batik Classification,” in 2024 International Electronics Symposium (IES), IEEE, Aug. 2024, pp. 449–453. doi: 10.1109/IES63037.2024.10665807.
  14. Q. W. ⋆, H. R. and J. L. Xiaofei Hui, “Class-Agnostic Object Counting with Text-to-Image Diffusion Model,” in European Computer Vision Association (ECVA),