GPR Image Analysis
Upload a GPR image for interface demonstration
Image Loaded Successfully
The GPR image is ready for model-based analysis. Connect this interface to the trained CNN inference endpoint to display actual prediction results.
An AI-assisted Ground-Penetrating Radar image analysis interface designed to study subsurface radar scan patterns and identify potential buried-object reflection candidates.
Landmine and IED Detection from Ground-Penetrating Radar Imagery
Ground-Penetrating Radar data is converted into standardized B-scan images for computer vision analysis.
A Convolutional Neural Network can be trained to learn visual patterns associated with different GPR image classes.
Image characteristics and model outputs can be examined to identify potential reflection candidates for further review.
Key goals of the proposed system
Convert radar output data into consistent 128 × 128 image representations.
Apply normalization and image-processing techniques before model analysis.
Develop a CNN-based image classification workflow for the prepared GPR dataset.
Present prediction results and relevant image information through a simple web interface.
From radar data to analysis
Radar signal acquisition
Normalization & resizing
Feature learning
Image classification
Candidate visualization
Upload a GPR image for interface demonstration
The GPR image is ready for model-based analysis. Connect this interface to the trained CNN inference endpoint to display actual prediction results.
Tools and technologies for the project
Ground-Penetrating Radar provides subsurface measurements that can be represented as radargrams or B-scan images. Computer vision methods can then be used to study patterns within these images.
This website is a project interface and research prototype. Final classification performance depends on the quality, labeling and training of the underlying GPR dataset and CNN model.