Bahrunani Punjabi (2025) GPR B-scan analysis with machine learning algorithms

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: UPCommons institutional repository (Universitat Politècnica de Catalunya)
• Year: 2025
• Date: 2025-09-01
• Authors: Bahrunani Punjabi, Mohit Bhagwan
• DOI: None

Research Groups

• Centre de Disseny d’Equips Industrials (CDEI)
• RootBot project

Short Summary

The study develops a noninvasive robotic system using Ground-Penetrating Radar (GPR) and the YOLO computer vision algorithm to detect subsurface objects, specifically tree roots, for improved agricultural water management.

Objective

• To identify and implement a reliable computer vision algorithm for real-time object detection in GPR data and evaluate the effects of data augmentation and post-processing on model performance.

Study Configuration

• Spatial Scale: Subsurface/Local (agricultural field scale)
• Temporal Scale: Real-time data analysis

Methodology and Data

• Models used: YOLO (You Only Look Once), GPRMax (for simulated data generation)
• Data sources: Field tests (tree root data), GPR scans from diverse applications, and simulated data; Roboflow was used for labeling and processing.

Main Results

• The trained YOLO model demonstrated satisfactory performance in detecting objects within GPR scans.

Contributions

• Integration of a noninvasive robotic GPR solution with a real-time object detection framework to assist in mapping root distribution for agricultural resilience.

Funding

• Not specified

Citation

@article{BahrunaniPunjabi2025GPR,
  author = {Bahrunani Punjabi, Mohit Bhagwan},
  title = {GPR B-scan analysis with machine learning algorithms},
  journal = {UPCommons institutional repository (Universitat Politècnica de Catalunya)},
  year = {2025},
  url = {https://openalex.org/W7111439074}
}