- Introduction
- What do they count?
- Where are they?
- When were they installed?
- Where can I access the data?
- Limitations
- Folder Structure
The City of Toronto employs Miovision Trafficlink Cameras to continuously monitor vehicle (including transit), bicycle, and pedestrian volumes at intersections across the city in an automated way. Miovision extracts volumes using Computer Vision AI from the camera footage and provides volume counts in aggregated 1 minute bins to the City via an API.
The program was initiated to support the King Street Transit Pilot by analysing the trends in volume on King Street, trends in volume on surrounding roads, and thru movement violations of the pilot. An example of how it was used to support the pilot project can be found here. Today the program continues to be expanded to monitor traffic at key locations around the city.
The data extracted from Miovision's API includes:
- Datetime aggregated to 1-minute bins
- Type of user (classification): Auto / Truck / Transit / Bicycle / Pedestrian
- Count of users
- Intersection entrance leg: N / S / E / W
- Type of (movement):
- Non-pedestrian: Through / Left / Right / U-Turn
- Pedestrians: Clockwise / Counterclockwise
You can see the locations of Miovision cameras on this map, noting the date_installed and date_decomissioned attributes for details on data availability.
- In 2017, 32 cameras were installed along King St and surrounding streets to monitor effects of the King Street Transit Pilot. 8 of these cameras were decommissioned in 2020.
- In 2020, 26 cameras were installed, primarily to monitor the effects of ActiveTO projects along Bloor, Danforth, University, Brimley, Huntingwood, and Faywood.
- In 2021, 10 additional cameras were installed, with 6 on Jane and Sheppard and 3 downtown along various bike projects.
- In fall 2023 and into winter 2024, a large expansion is underway, including 21 new cameras to create a cordon around downtown (generally Bathurst / Dupont / Don Valley), 10 new cameras to monitor traffic related to the SRT closure in Scarborough, and 4 new cameras along Bloor West.
Internally to the Transportation Data & Analytics team, data is stored in the bigdata RDS. For data automatically collected since January 2019, see the miovision_api schema and related documentation here. More limited data for the King Street Pilot from 2017 and 2018 can be found in the miovision_csv schema, which has a slightly different structure than the API data.
The data quality for the automatically processed API data stream (not including the King St Pilot study data from pre-2019 which was processed independently) is evolving, but the below rules are general guidelines for when the data became reliable to use:
- January 2019 for cars
- June 2019 for pedestrians
- Bike counts are still considered in development by Miovision in 2024 and should only used internally by D&A with extreme caution, or on analysis delivered by D&A. Any external sharing will be on a case by case basis depending on the purpose of the request.
Internal users can see additional detail on limitations for individual intersections and classifications in the miovision_api.anomalous_ranges table.
getting_started- This document contains an introduction for new users onUnderstanding Legs, Movement and Direction of Traveland some sample SQL queries.sql- Information on the database structure used for storing Miovision data and scripts for producing tables.api- Scripts and documentation for accessing the Miovision API. The Airflow DAGs are also described here.update_intersections- Contains up-to-date instructions for adding (or removing) new intersections to the database.csv- tracks baseline volumes and gaps.dev_notebooks- Jupyter Notebooks for development and testing.geojson- Contains a publicly accessible map of Miovision cameras.img- Contains images used in explanatory docs like this!Archive- This document contains information that has been deleted from the Miovision readmes.