Object classification in DataFromSky TrafficSurvey

Description and examples of object classes detected in TrafficSurvey

DataFromSky avatar
Written by DataFromSky
Updated over a week ago

DataFromSky TrafficSurvey platform can recognize up to 19 categories in your video. The categories are: Car, Van, Heavy Vehicle, Bus, Motorcycle, Bicycle, Pedestrian, Tuktuk, Light Truck, Scooter, Pram, Animal, Wheelchair, Tractor, Car Trailer, Truck Trailer, Caravan Trailer Caravan, and Tram. The objects are classified based on visual comparison with our wide datasets that follow the US FHWA Scheme, Standard UK Vehicle Classification Scheme, or European Scheme of vehicle classification. See the comparison of world classifications with DataFromSky categories at the end of this article!

Please note there are some differences in categorization between TrafficSurvey LIGHT and AERIAL. The difference is that AERIAL only recognizes the difference between the Medium vehicle (Van+Light vehicle grouped) and Heavy vehicle categories for cargo vehicles. However, the LIGHT TrafficSurvey can also differentiate these categories even further into Van, Light Truck, and Medium Truck categories.

Moreover LIGHT can additionally detect the Van, Tram, Scooter, Pram, Animal, Wheelchair, Tractor, Caravan, Caravan Trailer, Car Trailer and Truck Trailer

When needed you can also assign your own category in post-processing in the TrafficSurvey viewer for each object individually. If you have any questions don't hesitate to contact us using the live chat on the bottom right or send us an email here.

See the description and categorization of an object of each category below:

Categories common for LIGHT and AERIAL TrafficSurvey

Pedestrians
Stationary and moving people or roller-skaters.

Bicycle
Any type of pedal cycles including the person, e-scooters might be detected as Bicycle as well.

Motorcycle
Any type of motorcycle vehicle, scooters, or tuk-tuk including the driver.

Car
Sedan, Hatchback, MPV, MUV, SUV, Coupe, Pick-up, or any similar-size vehicle with usually 5 or fewer seats including those that pull light trailers. This category includes cars with an enclosed cab and an open cargo area with low sides as well.

Medium Vehicle - AERIAL (Light Truck for LIGHT analysis)
Medium size vehicle with luton body, such as cars for cargos, usually over 3,5t and less than 12t of weight. This category includes small, medium, large-size vans, large combi/crew vans, small motorhomes, smaller supply cars, or ambulances. 

Heavy Vehicle
Usually three and more axle truck, single-unit or multi-trailer, excavators. Agricultural or forestry tractor with their trailer is defined as a heavy vehicle as well.

Bus
Passengers-carrying large buses, school buses for more than 8 passengers. Tram might be identified as a bus as well.

Additional LIGHT TrafficSurvey categories (not present in AERIAL) - Van, Tram, Scooter, Pram, Animal, Wheelchair, Tractor, Caravan, Caravan Trailer, Car Trailer and Truck Trailer

Van (detected as Medium Vehicle in AERIAL analysis)

All van types including box-shaped vans.

Tractor

Mostly agricultural vehicles of varying shapes and sizes.

Tram

Currently focused on trams with above-head electrical wiring power supply.

Animal

Reliable detection of dogs and also detects larger animals like cats, horses, cattle, etc…

Caravan

Caravan trailers as well as caravan cars of varying sizes.

Caravan trailer

Detection of caravan trailers of various sizes and shapes.

Car Trailer

Small usually open trailers that can be towed behind passenger cars and smaller vehicles.

Truck Trailer

Heavy "industrial" trailers for shipping and transportation of larger quantities of goods and materials.

Pram (Carriage)

Various pram types. Note that the Pram generates 2 trajectories - the Pram and the Pedestrian pushing it.

Scooter

Includes both regular push scooters and e-scooters of varying sizes and wheel proportions. Note that the Scooter generates 2 trajectories - the Scooter and the Pedestrian riding it.

Wheelchair

Push wheelchairs and small wheelchair-type motorized carts for people with impaired mobility. Note that the Wheelchair generates 2 trajectories - the Wheelchair and the Pedestrian sitting in it.

Are you missing any category? Contact us! We can implement a new category detection for you! You can also assign a CUSTOM category to already detected objects. You can do so by selecting an object and clicking on Add Type... as shown in the image below. You can then assign this same category to all other existing objects.

How do categories fit classes according to different classification?
See mapping of DataFromSky Categories to international standards of vehicle classification, we have chosen three the most used: US FHWA Scheme, Standard UK Vehicle Classification Scheme, and European Scheme of vehicle classification.

US FHWA Scheme defines categories of Motorcycles that we do define as the category called "Motorcycle", Passenger Cars, Other Two-Axles, Four-Tire Single Unit Vehicles that are detected as a "Car" by DataFromSky system, Busses are labeled as a "Bus", and Two-Axle, Six-Tire, Single-Unit Trucks with Three-Axle Single-Unit Trucks are defined as a Medium Truck. Other larger vehicles are defined as a "Heavy Vehicle".

Standard UK Vehicle Classification Scheme defines Pedal Cycle, labeled as a "Bike" by DataFromSky Viewer, Cars and Light Goods Vehicle that is defined as a "Car", Public Service Vehicles defined as a "Bus", Ordinary Goods Vehicle 1 that is categorized as a "Medium Vehicle" and Ordinary Goods Vehicle 2 is categorized as a "Heavy Vehicle" by DataFromSky Viewer.

European Scheme divides vehicles into categories and sub-categories. Mapped DataFromSky Categories are the following: 

Have you already checked the recommendations for the resolution of the video and minimum size of objects defined separately for DataFromSky Light and DataFromSky Aerial that are necessary for accurate outputs? Or go through the list of data exports, you can get by processing the video through DataFromSky Viewer.

We will be happy to answer any questions you may have. Contact us!

Did this answer your question?