This repository contains the code to preprocess the spaceraft imagery dataset SPEED and train on it a detection neural network and a regression neural network. First of all, download the dataset:
wget https://stacks.stanford.edu/file/druid:dz692fn7184/speed.zip
unzip speed.zip
Then launch
python3 partition.py
To apply train/dev/test split to the dataset and
python3 labels.py
To create labels for both detection and regression network.
To the training, testing and export to OpenVINO format the passages in the two Jupyter Notebook can be followed.