Skip to content

Source code for PARKING OCCUPANCY ESTIMATION ON SENTINEL-1 IMAGES, ISPRS 2020

Notifications You must be signed in to change notification settings

sdrdis/s1_parking_occupancy

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PARKING OCCUPANCY ESTIMATION ON SENTINEL-1 IMAGES

Implementation for our ISPRS 2020 paper PARKING OCCUPANCY ESTIMATION ON SENTINEL-1 IMAGES.

Authors:

Step 1: generation of positions files

Positions can be generated from KMLs for example:

python utils_generate_positions_from_kmls.py kmls_folder id

Where kmls_folder is a folder containing kmls and id is an identifier (for example "retail_parkings")

Generates a NPZ file containing coordinates of all parkings to monitor in the positions folder.

For example, you can run the following command:

python utils_generate_positions_from_kmls.py sample_kml_folders test_parkings.npz

That will generate the positions/test_parkings.npz file.

This file can be opened with python:

import numpy as np
data = np.load('positions/test_parkings.npz')
data['shapes'] # contains all parkings coordinates
data['dirnames'] # contains filenames of kmls

Step 2: generate dataset

Prior to this step, we assume that you have created a sentinel-hub instance and that you have added the BOTH layer configured as shown below.

Image of Sentinel-Hub configuration

Generate dataset (example below from sentinel-hub website)

python generate_dataset_sentinelhub.py id sentinel_hub_instance_id

Where id is an identifier (for example "retail_parkings"), and sentinel_hub_instance_id is the Sentinel-hub instance id.

Generates folders in data/{id} containing images.

Example:

python generate_dataset_sentinelhub.py test_parkings {{YOUR_SENTINEL_HUB_ID}}

Will generate in the folder data/test_parkings all images data. The inside structure is the following:

  • parking_1
    • images # contains all images of parking 1
      • 20190101_060728_S1A_DESCENDING_008.tif
      • ...
    • metadata.npz # contains several informations about the parking (bounding box, shape of parking)
  • parking_2
    • ...

Step 3: evaluate masks

First way: simple mask

Generate parking masks.

python generate_masks.py

Generates images (all ids) in data indicating where the parking is and where it is not.

Second way: mask minus always occupied areas

Generate parking masks.

python generate_stats_per_orbit.py id
python generate_masks_minus_always_occupied.py id

Where id is an identifier (for example "retail_parkings").

generate_stats_per_orbit.py generates stats per orbit (median, std, min).

generate_masks_minus_always_occupied.py generates a mask removing areas where it is always occupied.

Step 4: evaluate occupancy

Estimate occupancy rates.

First way: simple thresholding

python estimate_simple.py

Applies a simple thresholding on all images (all ids). See parameters python estimate_simple.py --help.

Second way: thresholding compared to weekends

TODO

Step 5: generate csv

CSV are generated using following script:

python generate_csvs.py id

Where id is an identifier (for example "retail_parkings").

CSVs and graphs are generated in the output_csv/{id} folder.

About

Source code for PARKING OCCUPANCY ESTIMATION ON SENTINEL-1 IMAGES, ISPRS 2020

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages