A tool that searches the Mapillary database for images within a specific bounding box.

• Using Mapillary's API, we can further add features to narrow our search.
• Further, we can restrict the images to those uploaded within a certain date range (TODO)

This is an attempt to create an updated version of scripts used in geolocation photo retrieval automation, as these guides use defunct APIs like that of Wikimapia and Panoramio.

• Unfortunately, Mapillary appears to be mostly consisting of crowd-source images of roads.

Ensure you are within the mapillary-search directory.

pip install -r requirements.txt

Next, obtain an access token from Mapillary via this link: https://www.mapillary.com/dashboard/developers

Replace line 8 of mapillary_search.py with the Client Token:

# API Details
#TODO: Replace API key below
API_KEY = "MLY|123412341234|123412341234"

python3 mapillary_search.py southwest northeast [-l limit] [-d detections]

• southwest: The south-west corner of your bounding box, expressed as Latitude, Longitude within quotes. Example: "1.3518580962691118,103.93356798061131". No, I don't stay at these coordinates.
• northeast: The north-east corner of your bounding box, expressed as Latitude, Longitude within quotes. Example: "1.3552656368803468,103.93720991127293"
• limit: The maximum number of images to be extracted, ignoring images that have been filtered out by detections.
• detections: Features of the image to further narrow the search,
  • See this link for a full list of possible detections. Note that our tool does not support the detections that are unavailable for bounding box search.
  • This uses Mapillary's API to extract features of the image such as signs, trash-cans.
  • When entering the detections, enter the value of the detection, separated by commas without spaces.
  • Example: object--sign--advertisement,object--trash-can to search for images with either Advertisement Signs or Trash Cans.

The tool will then generate a HTML file with every extracted image. Clicking on the image links to an OpenStreetMap URL with a marker on the exact coordinates of the image.

Input:

py .\mapillary_search.py "1.3518580962691118,103.93356798061131" "1.3552656368803468,103.93720991127293" -l 10 -d object--sign--advertisement,object--trash-can

Output:

[*] Saving to log file...
[*] Saved 5 images to: mapillary-search\log\Log_2023-12-23_09-09-05.html

• Implement the date range filter.