Publications
Journal Articles
2023
1.
Nairit Barkataki; Banty Tiru; Utpal Sarma
Size estimation of underground targets from GPR frequency spectra: A deep learning approach Journal Article
In: Journal of Applied Geophysics, vol. 213, pp. 105009, 2023, ISSN: 0926-9851.
Abstract | BibTeX | Tags: Archeology, Artificial Neural Network, Bridge inspection, Civil Engineering, Deep learning, Ground penetrating radar, neural networks, object size prediction | Links:
@article{barkataki2023size,
title = {Size estimation of underground targets from GPR frequency spectra: A deep learning approach},
author = {Nairit Barkataki and Banty Tiru and Utpal Sarma},
url = {https://nairit.in/wp-content/uploads/Paper009_JAG_ObjectSizeAScan.pdf},
doi = {10.1016/j.jappgeo.2023.105009},
issn = {0926-9851},
year = {2023},
date = {2023-06-01},
urldate = {2023-06-01},
journal = {Journal of Applied Geophysics},
volume = {213},
pages = {105009},
abstract = {GPR (Ground Penetrating Radar) is a robust and effective device for identifying underground artefacts. Construction companies and civil engineers should be aware of the sizes of rebars and pipelines before and during construction work for various reasons. Most research efforts have typically concentrated on GPR signal analysis in the time domain, however recent studies have increasingly focused on analysis in the frequency domain. This paper proposes an artificial neural network (ANN) model for estimating the diameter of underground rods (solid) and pipes (hollow). GPR data captured in the time domain domain is transformed to the frequency domain using FFT, after which feature extraction is performed using ANN. An FPGA-based prototype GPR system is used to collect GPR A-Scan data for a variety of targets made of aluminium, stainless steel, rebar, and PVC. A mean absolute percentage error of 1.89% is achieved using the proposed model. The experimental results confirm the effectiveness of the proposed approach in extracting size-related information from GPR data.},
keywords = {Archeology, Artificial Neural Network, Bridge inspection, Civil Engineering, Deep learning, Ground penetrating radar, neural networks, object size prediction},
pubstate = {published},
tppubtype = {article}
}
GPR (Ground Penetrating Radar) is a robust and effective device for identifying underground artefacts. Construction companies and civil engineers should be aware of the sizes of rebars and pipelines before and during construction work for various reasons. Most research efforts have typically concentrated on GPR signal analysis in the time domain, however recent studies have increasingly focused on analysis in the frequency domain. This paper proposes an artificial neural network (ANN) model for estimating the diameter of underground rods (solid) and pipes (hollow). GPR data captured in the time domain domain is transformed to the frequency domain using FFT, after which feature extraction is performed using ANN. An FPGA-based prototype GPR system is used to collect GPR A-Scan data for a variety of targets made of aluminium, stainless steel, rebar, and PVC. A mean absolute percentage error of 1.89% is achieved using the proposed model. The experimental results confirm the effectiveness of the proposed approach in extracting size-related information from GPR data.
Conference Proceedings
2022
1.
Nairit Barkataki; Ankur Jyoti Kalita; Utpal Sarma
Automatic Material Classification of Targets from GPR Data using Artificial Neural Networks Proceedings Article
In: 2022 IEEE Silchar Subsection Conference (SILCON), IEEE 2022, ISBN: 978-1-6654-7100-8.
Abstract | BibTeX | Tags: Archeology, Artificial Neural Network, classification, Deep learning, Ground penetrating radar, Landmine detection, Object Material | Links:
@inproceedings{barkataki2022automatic,
title = {Automatic Material Classification of Targets from GPR Data using Artificial Neural Networks},
author = {Nairit Barkataki and Ankur Jyoti Kalita and Utpal Sarma},
url = {https://nairit.in/wp-content/uploads/2022_SILCON_Barkataki_Material.pdf},
doi = {10.1109/SILCON55242.2022.10028944},
isbn = {978-1-6654-7100-8},
year = {2022},
date = {2022-11-06},
urldate = {2022-11-06},
booktitle = {2022 IEEE Silchar Subsection Conference (SILCON)},
journal = {2022 IEEE Silchar Subsection Conference (SILCON)},
organization = {IEEE},
abstract = {Ground penetrating radar (GPR) is a preferred non-destructive method to study and identify buried objects in the field of geology, civil engineering, archaeology, military, etc. Landmines are now largely composed of plastic and other non-metallic materials, while archaeologists must deal with buried artefacts such as ceramics, pillars, and walls built of a range of materials. As a result, understanding the material properties of buried artefacts is critical. This study presents an ANN model for automatic classification of buried objects from GPR A-Scan data. The proposed ANN model is trained and validated using a synthetic dataset generated using gprMax. The model performs well in classifying three different object classes of aluminium, iron and limestone, while achieving an overall accuracy of 95%.},
keywords = {Archeology, Artificial Neural Network, classification, Deep learning, Ground penetrating radar, Landmine detection, Object Material},
pubstate = {published},
tppubtype = {inproceedings}
}
Ground penetrating radar (GPR) is a preferred non-destructive method to study and identify buried objects in the field of geology, civil engineering, archaeology, military, etc. Landmines are now largely composed of plastic and other non-metallic materials, while archaeologists must deal with buried artefacts such as ceramics, pillars, and walls built of a range of materials. As a result, understanding the material properties of buried artefacts is critical. This study presents an ANN model for automatic classification of buried objects from GPR A-Scan data. The proposed ANN model is trained and validated using a synthetic dataset generated using gprMax. The model performs well in classifying three different object classes of aluminium, iron and limestone, while achieving an overall accuracy of 95%.