Deep learning capabilities with Mobotix high-end cameras

Stream2IQ capabilities

Persons recognition

Face, attitude, expression, posture

Objects recognition

Accurate objects detection

Complex scenes analysis

Events, public places, stores

Deep learning outstanding features

  • "Understand" images from live or recorded video streams
  • Classify and prioritize objects
  • Sequence algorithms according to image content
  • LAPI recognition for vehicules, type, color, brand, plate number
  • Persons facial , attitude and expression recognition
  • Create indexed databases including metadata
  • Create search rules
  • Create scenarios to analyse data and issue statistics

Deep learning vs machine learning

Machine learning adapts algorithm but needs an expert to extract data

Deep learning extracts information and adapts algorithm

Real time automatic objects identification

Scene understanding

"I see a man on a horse near a dog, his car is parked under a tree. Another person looks at them and seems happy."

 

 

Video analysis capabilities

  • Analyse content of all images one by one

  • Detect, classify, segment objects with highest accuracy and lowest error rate

  • Capability to dynamically adjust error margin depending on light, distance and other parameters

  • Detect intrusion of specific objects while other others are considered as accepted

  • Count objects according to their position and movement

  • Perform specific data analysis for vehicules

    • Counting by type, brand, model
    • Counting by direction and movement
    • Detection of stopped or parked car in forbiden location
  • Perform specific data analysis for persons

    • Counting
    • Attitude, posture, expression, facial recognition
  • Index images with metadata to intuitively seek for recorded sequences

Search for specific scenes within recorded database metadata

Search for all white vans with ID plate starting by FA

Person riding bicycle with a dog

Waiting line above 10 persons


Discuss your projects with us