Skip to content

Instantly share code, notes, and snippets.

@cbeddow
Last active September 16, 2024 14:02
Show Gist options
  • Save cbeddow/28e5d043a46ba34ea91f7b66564307d4 to your computer and use it in GitHub Desktop.
Save cbeddow/28e5d043a46ba34ea91f7b66564307d4 to your computer and use it in GitHub Desktop.
Python script for downloading data from Mapillary API v4 in a bounding box
import mercantile, mapbox_vector_tile, requests, json
from vt2geojson.tools import vt_bytes_to_geojson
# define an empty geojson as output
output= { "type": "FeatureCollection", "features": [] }
# vector tile endpoints -- change this in the API request to reference the correct endpoint
tile_points = 'mly_map_feature_point'
tile_traffic_signs = 'mly_map_feature_traffic_sign'
tile_coverage = 'mly1_public'
# tile layer depends which vector tile endpoints:
# 1. if map features or traffic signs, it will be "point" or "traffic_sign" respectively
# 2. if looking for coverage, it will be "image" for points, "sequence" for lines, or "overview" for far zoom
tile_layer = "point"
# Mapillary access token -- user should provide their own
access_token = 'MLY|XXX'
# a bounding box in [west_lng,_south_lat,east_lng,north_lat] format
west, south, east, north = [-80.13423442840576,25.77376933762778,-80.1264238357544,25.788608487732198]
# list of values to filter for and keep -- update this if changing to traffic signs
filter_values = ['object--support--utility-pole','object--street-light']
# get the list of tiles with x and y coordinates which intersect our bounding box
# MUST be at zoom level 14 where the data is available, other zooms currently not supported
tiles = list(mercantile.tiles(west, south, east, north, 14))
# loop through list of tiles to get tile z/x/y to plug in to Mapillary endpoints and make request
for tile in tiles:
tile_url = 'https://tiles.mapillary.com/maps/vtp/{}/2/{}/{}/{}?access_token={}'.format(tile_points,tile.z,tile.x,tile.y,access_token)
response = requests.get(tile_url)
data = vt_bytes_to_geojson(response.content, tile.x, tile.y, tile.z,layer=tile_layer)
# filter for only features matching the values in filter list above
filtered_data = [feature for feature in data['features'] if feature['properties']['value'] in filter_values]
# check if features are inside bbox, and push to output geojson object if yes
for feature in filtered_data:
if (feature['geometry']['coordinates'][0] > east and feature['geometry']['coordinates'][0] < west)\
and (feature['geometry']['coordinates'][1] > south and feature['geometry']['coordinates'][1] < north):
output['features'].append(feature)
# save a local geojson with the filtered data
with open('mydata.geojson', 'w') as f:
json.dump(output, f)
@jyothir07
Copy link

@cbeddow the above code is not working. Can u please suggest

@tordans
Copy link

tordans commented Aug 27, 2024

FYI I learned recently, that the number of image points in the vector tiles at https://tiles.mapillary.com are limited to about 2k per tile. Depending on the use case, it is not the right approach to fetch the data. The mapillary JSON api does have a 2k limit as well but there are plans to add pagination to those endpoints which will make it easier to fetch all data for a given reagion+filter.

@cbeddow
Copy link
Author

cbeddow commented Sep 16, 2024

@jyothir07 can you share the errors you get?

@tordans yes this may end up working better. I also have done it before where I split into like zoom 16 tiles, then get a tile bbox, and just continuous make a request to the JSON API with that bbox so it's always small but in theory it could still exceed 2000, though typically it's because of someone uploading 2000+ images with the same GPS point

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment