YOLO-SOD: Color probability map-based fuzzy framework using deep models for face detection
- 1Department of Electronic Engineering, Faculty of Electrical and Computer Engineering.NED University of Engineering & Technology, Karachi, Pakistan.
- 2Department of Computer Science & Information Technology, Faculty of Electrical and Computer Engineering.NED University of Engineering & Technology, Karachi, Pakistan.
- 3Institute for Advanced Technology Convergence (Mobility), Kyungpook National University, Buk-gu, Daegu, Republic of Korea.
Abstract
Detecting small faces remains challenging due to scale variation and limited discriminative cues at low resolutions. We address this with a two-pronged approach. First, we modify YOLOv5 by adding extra convolutional layers and small-object–focused detection heads that operate on larger, higher-resolution backbone feature maps, improving sensitivity to tiny targets. Second, we introduce a dynamic, image adaptive patching strategy for scale selection. We compute Cr channel color probability maps and measure the number and density of pixel clusters; a Fuzzy Inference System (FIS) uses these cues to select the most suitable patch size for near and far faces. On the WIDER FACE benchmark, our method attains 93.07% mAP@50 (Easy), 91.25% mAP@50 (Medium), and 84.45% mAP@50 (Hard). On our proposed local dataset, the proposed model achieves 49.36% mAP@50 for different size faces. Beyond faces, the approach generalizes to small object detection on VisDrone-DET2019, reaching 46.76 mAP@50. These results demonstrate that coupling small-object-aware architecture design with color-guided, fuzzy patch selection yields competitive mAP@50 for small and distant targets.
Related
- BCG-YOLO: an efficient YOLO-based framework for multi-class blood cell detection in microscopic images.
- A lightweight YOLO based framework for real-time seat belt detection in intelligent traffic surveillance
- Cracks-YOLO: an improved YOLOv5 tongue crack detection method based on switchable atrous convolution and transformer
- DBRA-SCSA-YOLO: a multi-attention network for image-based student classroom behavior detection
- LHR-YOLO: A SAR small-target ship detection method based on improved YOLO11
BibTeX
@article{saba2026yolosod,
title = {{YOLO-SOD: Color probability map-based fuzzy framework using deep models for face detection}},
author = {Aniqa Saba and Muhammad Faizan Shirazi and Yawar Rehman and Muhammad Mubashir Khan},
journal = {PLOS One},
year = {2026},
month = oct,
doi = {10.1371/journal.pone.0360794},
url = {https://doi.org/10.1371/journal.pone.0360794},
note = {CC BY},
}
RIS
TY - JOUR
TI - YOLO-SOD: Color probability map-based fuzzy framework using deep models for face detection
AU - Aniqa Saba
AU - Muhammad Faizan Shirazi
AU - Yawar Rehman
AU - Muhammad Mubashir Khan
PY - 2026
DA - 2026/10/22
JO - PLOS One
DO - 10.1371/journal.pone.0360794
UR - https://doi.org/10.1371/journal.pone.0360794
AB - Detecting small faces remains challenging due to scale variation and limited discriminative cues at low resolutions. We address this with a two-pronged approach. First, we modify YOLOv5 by adding extra convolutional layers and small-object–focused detection heads that operate on larger, higher-resolution backbone feature maps, improving sensitivity to tiny targets. Second, we introduce a dynamic, image adaptive patching strategy for scale selection. We compute Cr channel color probability maps and measure the number and density of pixel clusters; a Fuzzy Inference System (FIS) uses these cues to select the most suitable patch size for near and far faces. On the WIDER FACE benchmark, our method attains 93.07% mAP@50 (Easy), 91.25% mAP@50 (Medium), and 84.45% mAP@50 (Hard). On our proposed local dataset, the proposed model achieves 49.36% mAP@50 for different size faces. Beyond faces, the approach generalizes to small object detection on VisDrone-DET2019, reaching 46.76 mAP@50. These results demonstrate that coupling small-object-aware architecture design with color-guided, fuzzy patch selection yields competitive mAP@50 for small and distant targets.
SN - 1932-6203
ER -