Ground Based SAR Data for Classification - 9 Objects in the Near Distance

GBSAR parameters
The dataset consists of 267 raw Ground Based SAR (GBSAR) measurements obtained with GBSAR-Pi [1].
GBSAR-Pi works in stop-and-go mode. GBSAR sensor is FMCW (Frequency Modulated Continuous Wave) radar with central frequency at 24 GHz. Total GBSAR aperture is 20 cm with 0.5 cm step size (40 steps). FMCW radar periodically transmits sawtooth signal every 155 ms. Wave polarization of FMCW radar is set to vertical.

Classes
There are 4 aluminum objects, 3 glass, and 2 plastic objects with different shapes and sizes. The objects were distanced approximately 30 cm from the radar.

Naming:
Each of the test objects used in the measurements represent one class. Therefore, there are 9 classes in total. Class naming is based on the material of the object and its size in a way that A4 is bigger than A1, but both of them are made from aluminum.

Number of data points per class:
A1 - 27
A2 - 29
A3 - 28
A4 - 30
G1 - 35
G2 - 33
G3 - 29
P1 - 28
P2 - 28

Data format
GBSAR matrices are given in .npy format. The dimensions of the matrices are 40x1024 (40 GBSAR steps x 1024 FMCW frequency points).

This work was supported in part by Croatian Science Foundation (HRZZ) under the project number IP-2019-04-1064.

[1] Kačan M, Turčinović F, Bojanjac D, Bosiljevac M. Deep Learning Approach for Object Classification on Raw and Reconstructed GBSAR Data. Remote Sensing. 2022; 14(22):5673. https://doi.org/10.3390/rs14225673