Dataset and code for "OxiMoRa: Oxidation Monitoring of Electrochemical Steel Corrosion based on Harmonic Radar and Deep Learning"

Dataset and code for "OxiMoRa: Oxidation Monitoring of Electrochemical Steel Corrosion based on Harmonic Radar and Deep Learning". IQ samples collected with harmonic radar and converted to grayscale images for computer vision classification. Varying quantities of corroded items (Screws) were displayed in front of the SDR-based radar, and their electromagnetic fingerprints were recorded.

OxiMoRa-main.zip contains all the code for reproducibility of the results.

The rest of the dataset has been split into subfiles for upload.

Scenario 1 - Detection has a bag of rusty screws displayed in front of a radar:

--> noise_f25_1 to noise_f25_4

--> rustbag_f25_1 to rustbag_f25_4

Scenario 2- Screws amount has different recordings with a different number of rusty screws in front of a radar. The purpose is to see if the radar can quantify to some extent the corrosion severity.

--> 5 classes, one for each number of screws, e.g., 4 screws --> 4screw_1 to 4screw_3

Scenario 3 has different settings based on pipe testing: 

Pipe_corroded are files with rusty screws inside to mimic internal corrosion.

--> Pipe_corroded_1 to Pipe_corroded_4

Pipe_empty are files for baseline knowledge of the pipe.

--> Pipe_empty_f25_1 to Pipe_empty_f25_6

Pipe_normal_nails are files where healthy screws are introduced in the pipe (galvanized steel for both the screws and the pipe)

--> Pipe_healthy_nailsf25_1 to Pipe_health_nailsf25_4