Bermudo et al. (2026) Performance Analysis of YOLO-Based Architecture for Water Level Monitoring

Identification

• Journal: Lecture notes in networks and systems
• Year: 2026
• Date: 2026-01-01
• Authors: Mary Ann Gliefen A. Bermudo, Apple Rose B. Alce, Adrian P. Galido
• DOI: 10.1007/978-3-032-10824-1_28

Research Groups

• Mindanao State University - Iligan Institute of Technology, Iligan City, Philippines

Short Summary

This paper evaluates the performance of YOLOv5 and YOLOv8 models for identifying and classifying river water levels using an image-based dataset. The study found that larger architectures, specifically YOLOv5xu and YOLOv8x, achieved the highest mean Average Precision, demonstrating their effectiveness for continuous water level monitoring.

Objective

• To assess the effectiveness of YOLOv5 and YOLOv8 models in identifying and classifying river water levels using an image-based dataset.

Study Configuration

• Spatial Scale: River segments and riverbanks, focusing on areas prone to fluctuating water levels and floods.
• Temporal Scale: Designed for continuous monitoring and real-time data acquisition.

Methodology and Data

• Models used: YOLOv5 (YOLOv5xu, YOLOv5nu) and YOLOv8 (YOLOv8x, YOLOv8n) object detection models.
• Data sources: Image-based dataset, Roboflow datasets, Ultralytics Hub. Training and testing were conducted using Google Colab.

Main Results

• YOLOv5xu and YOLOv8x achieved the highest mean Average Precision (mAP) of 82.1%.
• YOLOv8x demonstrated an accuracy of 0.965 and a recall of 0.936.
• YOLOv5xu showed a precision of 0.964 and a recall of 0.939.
• Larger and more complex architectures (YOLOv5xu, YOLOv8x) consistently outperformed their simpler counterparts.
• Simpler models, YOLOv5nu and YOLOv8n, exhibited lower performance with mAPs of 53.5% and 56.5%, respectively.

Contributions

• Demonstrated the high accuracy, precision, and recall of YOLO-based models, particularly YOLOv5xu and YOLOv8x, for detecting river water levels.
• Provided reliable data suitable for infrastructure planning and decision-making processes in flood-prone areas.
• Proposed the integration of these models with IoT-based monitoring systems to enhance real-time data acquisition and improve responsiveness to environmental changes.

Funding

• Not specified in the provided text.

Citation

@article{Bermudo2026Performance,
  author = {Bermudo, Mary Ann Gliefen A. and Alce, Apple Rose B. and Galido, Adrian P.},
  title = {Performance Analysis of YOLO-Based Architecture for Water Level Monitoring},
  journal = {Lecture notes in networks and systems},
  year = {2026},
  doi = {10.1007/978-3-032-10824-1_28},
  url = {https://doi.org/10.1007/978-3-032-10824-1_28}
}