Starling CV is a pedestrian and vehicle trajectory prediction framework, powered by a neural network, that helps improve road safety.

In Starling CV a neural network and computer vision system monitors and classifies moving objects in busy road and roadside scenes in real-time, calculates their trajectories, and infers the location of hidden pedestrians or cyclists (e.g. behind high-sided vehicles and buses). This enables the system to detect, predict and respond to changing safety conditions on pavements, roads and crossings.

The system can be installed at any road junction or crossing, and is used to control interactive road and pavement surfaces as well as curbside lighting and road markings, making pedestrians, cyclists & drivers safer and more aware of each other.

Overview of Starling CV


  • Using a neural network framework, Starling CV tracks objects moving across road and pavement surfaces
  • Distinguishes between different types of pedestrians, cyclists and vehicles
  • Calculates precise locations, trajectories and velocities
  • Predicts near-future paths and speeds of moving objects
  • Evaluates probability of various safety and danger conditions
  • Optional: interactive road surface
  • Optional: Design framework for interactive road, pavement and crossing patterns

Neural network framework tracks objects moving in Starling CV
Neural network framework tracks objects moving in Starling CV

Expected Outcomes

By deploying this system, it will:

  • Improve safety relationships between people, cyclists and vehicles
  • Generate valuable information about a site’s pavement and road usage
  • Road markings and lighting can adapt to different usages for different times of day or different conditions (e.g. street markets or events) and respond in different ways depending on current usage

Use Cases

  • Pedestrian and vehicular flow analysis and prediction
  • Dynamic pavement zoning
  • Dynamic curb lighting for delivery and driver coordination