Vision system designed for automotive to provide virtual camera perspective from four fisheye cameras mounted on vehicle sides.

Sample inputs

Sample outputs

Keep in mind, that it is not a final product. Some things may be unnecessary and not written in the most optimal form.

Please edit the environment.yml, so that the last (72) line points to the location of the new enviroment. More about creating an enviroment from an yml file.

To use the spec file to create an identical environment:

conda env create -f environment.yml

OR

To use the spec file to create an identical environment:

conda create --name FreeCameraForAutomotive --file requirements.txt

To use the spec file to install its listed packages into an existing environment:

conda install --name FreeCameraForAutomotive --file requirements.txt

OR

conda create --channel conda-forge --name FreeCameraForAutomotive python=3 pillow vispy=0.6 opencv=3.4 pyqt

It is possible to swap [PyQt5] for any of following packages: ['PyQt4', 'PyQt5', 'PySide', 'Pyglet', 'Glfw', 'SDL2', 'wx', 'EGL', 'osmesa'].

To run simply type python Video_Viewer.py in console while in current directory. Or use Video_Viewer.spec with PyInstaller to create .exe

conda install -c conda-forge pyinstaller
pyinstaller yourprogram.spec

Running last line with additional parameters --noconfirm will skip confirming decision and --onefile will generate single .exe file instead of whole directory.

By default, the video input is 4 independent .mp4 video files, which can be changed in the [load_cameras] function in the [Video_Processor] class. In other cases, changing the [CAMERA_READ_FROM_FILE] flag to False will capture the video stream.

CarModel.obj is a free model with Personal Use License downloaded from [https://free3d.com/3d-model/automobile-v1--84248.html]. And datasets were rendered using The "Multi-FoV" synthetic datasets, which was released under the Creative Commons license (CC BY-NC-SA 3.0), which is free for non-commercial use (including research).