CARLA is an open-source autonomous vehicle driving suite built on top of Unreal Engine 4 which provides a Python API to manage the simulations. In this repo, I built a data collector on top of it, which spawns vehicles, pedestrians, changes the weather to a custom set, and allocates sensors (RGB, 2D bounding box and depth) on random vehicles to record all data into a single HDF5 file.
With the data recorded from this repo, one can quickly build their own private dataset to train
machine learning models for autonomous driving and other applications ![]()
Take a peek into the source code