Notebooks

Some notebooks are available in the notebooks directory of the CARS project. CARS pipeline notebook can be used to launch CARS prepare and compute_dsm in a notebook Following specific notebooks can be used to compute intermediary results and statistics using the CARS API.

Notebooks preparation

CARS has to be installed and a jupyter notebook configuration has be set up.

Quick local install

After installing Dependencies, use the following automated make command line:

make notebook

Advanced install

The automated make notebook command line installs CARS with notebook dependencies:

$ make install-deps        # Install venv virtualenv and CARS dependencies
$ source venv/bin/activate # Go in the virtualenv
$ pip install .[notebook]  # Install CARS with notebook dependencies
or
$ pip install .             # Install CARS standalone
$ pip install jupyter bokeh # Install notebook dependencies

Then, a Jupyter kernel can be created in the virtualenv with the following command:

$ python -m ipykernel install  --sys-prefix --name=cars-venv

Finally, launch a local jupyter notebook environment with:

$ jupyter notebook

CARS pipelines notebook

The cars_pipelines.ipynb notebook show a complete CARS 3D run. From CARS demo data, it executes the prepare and compute_dsm pipelines from pipeline API.

Warning

The whole CARS kernel described before is needed to launch this notebook.

It depends and includes cars_generic.ipynb which contains generic functions, demo data configuration. This generic playbook have to be on the same directory than cars_pipelines.ipynb notebook.

Specific Notebooks

Warning

The following notebooks require CARS generated pipeline’s outputs (prepare or compute_dsm). These outputs directories have to be inserted in the notebooks beginning configuration.

Step by step compute DSM

The step_by_step_compute_dsm.ipynb notebook explains how to run step by step DSM computation with CARS, starting from the prepare step ouptut folder.

The following parameters have to be set :

cars_home : Path to the CARS folder.
content_dir : Path to the directory containing the content.json file of the prepare step output.
roi_file : ROI to process from a file size. It can be either a path to a vector file or a raster file as expected by CARS. Put roi_file=None to use roi_bbox (mutually exclusive and roi_file has precedence ).
roi_bbox : ROI bounding box of 4 floats to process if roi_file is not defined (None). roi_bbox = ["xmin", "ymin", "xmax", "ymax"]. They are considered in in final projection depending on EPSG code.
output_dir : Path to output dir where to save figures and data.

Epipolar distributions

The epipolar_distributions.ipynb notebook enables to visualize the distributions of the epipolar error and disparity estimated from the matches computed in the preparation step.

The following parameters have to be set :

cars_home : Path to the CARS folder.
content_dir : Path to the directory containing the content.json file of the prepare step output.

Low resolution DSM fitting

The lowres_dem_fit.ipynb notebook details how to estimate and apply the transform to fit a DSM to the low resolution initial DEM. This method is currently implemented in CARS.

The following parameters have to be set :

cars_home : Path to the CARS folder.
content_dir : Path to the directory containing the content.json file of the prepare step output.

Compute DSM memory monitoring

The compute_dsm_memory_monitoring.ipynb notebook shows how to load data and plot graph to monitor memory consumption during execution of CARS compute_dsm step with Dask.

The following parameters have to be set :

compute_dsm_output_dir : The output folder of the compute DSM step
nb_workers_per_pbs_jobs (Optional) : The number of workers process per pbs job (default : 2)
nb_pbs_jobs : The number of pbs jobs (Number of workers divided by nb_workers_per_pbs_jobs)