Skip to main content

Datasets

Standard Dataset

FDFD Simulations

Citation Author(s):
Mahshid Asri (Northeastern University)
Ann Morgenthaler (Northeastern University)
Carey Rappaport (Northeastern University)
Submitted by:
Mahshid Asri
Last updated:
DOI:
10.21227/r8td-a279
Data Format:
No Ratings Yet

Abstract

This paper presents a deep learning model for fast and accurate radar detection and pixel-level localization of large concealed metallic weapons on pedestrians walking along a sidewalk. The considered radar is stationary, with a multi-beam antenna operating at 30 GHz with 6 GHz bandwidth. A large modeled data set has been generated by running 2155 2D-FDFD simulations of torso cross sections of persons walking toward the radar in various scenarios. 

Instructions:

This data includes two folders. The folder "Images" contains the reconstructed radar images for all the simulated case. Each simulated case contains 4 separate images:

image_casenumber_3.png corresponds to the reconstruced image when the radar focuses on the left arm

image_casenumber_2.png corresponds to the reconstruced image when the radar focuses on the right arm

image_casenumber_1.png corresponds to the reconstruced image when the radar focuses on the middle of the torso

image_casenumber_0.png is generated by combining the images above in a complex format (using their amplitude and phase).

 The folder "Masks" contains the ground truth files used for each simulation. The 4 ground truth images corresponding to each case are exactly alike and are added for making the training part easier. 

Funding Agency
U.S. Department of Homeland Security, Science and Technology Directorate
Grant Number
22STESE00001-02-00