Make and Model Recognition (MMR)

Highlights

16 categories
308 makes
2521 models
6658 generations
14 TAGs
front + rear + side views
day + night light

Accuracy & Performance

99% category
98% make and model
95% generation & variation
4 ms GPU speed

Make and model recognition is an AI-based software providing detailed vehicle classification. The product recognizes with extremely high accuracy the category, make, model, generation, variant and color of the captured cars from both front and rear views in day and night conditions. 

A unique feature of the product is the recognition of so-called TAGs, i.e. detection of emergency vehicles, fire brigade, pickup truck, taxis... the software can distinguish 14 such tags. 

It also recognizes objects from the field of softmobility, i.e. it is able to detect pedestrians, cyclists, animals, strollers and other trolleys...  

Each new version of MMR is always updated with newly launched car models based on data from Europe, North America, East Asia, the Middle East, Africa and Australia. 

Typical applications: toll systems, traffic safety and surveillance systems, law enforcement, entrance and parking systems, police and governmental applications, traffic infrastructure planning systems, smart cities and others.

TAG recognition

ambulance
law enforcement
fire brigade
caravan
pickup
push bumper
rear mount
wood truck
animal transport
taxi
towed
tank truck
mixer truck
rickshaw

New cloud service

  • The launch of a cloud service based on our MMR (car manufacturer and model recognition) and ANPR (license plate detection and reading) analytics, available at https://www.trafficeye.ai. You can now test the operation of our unique technologies online. The processing of the first 50 images is free for each user. 

  • Massive improvement in MMR accuracy for the US region. 

Feature Product type
SDK Embedded SDK REST API Server
Recognition Category + Make + Model + Generation + Variation + Colour + Tags
Camera View Frontal + Rear
Regions Europe, Middle East, North America, Asia&Pacific, South Africa
Target Implementation Cameras & Computer connected cameras Customer Proprietary Applications Service on Demand at client site
Supported Platforms
  • Windows (Win32 & x64)
  • Unix (x86_64 & AArch64)
  • CPU / GPU
  • *other on request
  • Windows (Win32 & x64)
  • Unix (x86_64 & AArch64)
  • CPU / GPU
  • *other on request
  • Docker image
  • Ubuntu x86_64
  • CPU / GPU
  • *other on request
    License type Time unlimited license with 12 months support (extendable)
    Performance 4 ms per vehicle on GPU, 8 ms on CPU, see technical sheet for details
    Accuracy
  • Category > 99 %
  • Make and model > 98 %
  • Generation and variation > 95 %
  • Colour > 80 %
  • *in normal daylight conditions
    Online Demo Yes, available here
    *you can also request offline demo

    Try our demo on trafficeye.ai

    In action

    Contact us

    Our references