SORT<T>

Models & Types · AiDotNet.ComputerVision.Tracking

SORT (Simple Online and Realtime Tracking) implementation.

Inheritance: ObjectTrackerBase<T> → SORT<T>

For Beginners

SORT is a simple yet effective online tracking algorithm that uses Kalman filtering for state estimation and the Hungarian algorithm for detection-to-track association based on IoU overlap.

How It Works

This follows the reference implementation (abewley/sort): a 7-state constant-velocity Kalman filter over (u, v, s, r) with filterpy’s noise settings, an optimal assignment on the IoU matrix, matches below IouThreshold rejected, and a track reported only in frames where it was matched and its hit streak has reached MinHits (or during the first MinHits frames).

Key features:

  • Kalman filter for motion prediction
  • IoU-based association using Hungarian algorithm
  • No appearance features (pure motion-based)
  • Real-time performance (~260 FPS)

Reference: Bewley et al., “Simple Online and Realtime Tracking”, ICIP 2016

Constructors

Constructor Summary
SORT(TrackingOptions<T> options) Creates a new SORT tracker.

Properties

Property Summary
string Name { get; }

Methods

Method Summary
TrackingResult<T> Update(List<Detection<T>> detections)
void Reset()