Raw Radar Data Generation

---

This repository is the the code used in the paper: Generation of Realistic Synthetic Raw Radar Data for Automated Driving Applications using Generative Adversarial Networks. Above there a summary of this method.

Instalation

---

$ git clone <repository>

Accessing the Dataset

The dataset file EXP_17_M.h5 is also avaiable on kaggle.

Usage

---

Preprocessing

In order to perform the training is necessary preprocess the data, that are in h5 file. For this run all the jupyter netebook:

save_prerpocessed_chirps_labels.ipynb

This notebook will parse the data to numpy array and apply the change of scales. At the end of the execution will have one .npy file with the chirps data and other with the labels.

Training the model

Use thte follow script to train the model.

python src/models/trainings/train_conditional.py

This code uses the Weights & Bias API.

Saving the generations

To save the generations in a numpy file.

python src/generate/save_conditional.py

Saving RA maps

To save RA maps as png files.

python src/generate/save_ramaps.py