Video Object Re-Identification Thresholds for Scene Exit Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional re-identification methods in object tracking systems risk incorrectly identifying similar but different objects as the same due to visually similar appearances, especially in scenarios where objects exit and re-enter the scene, leading to loss of tracking information.

Innovation Solution

Adjust the re-identification threshold to require greater similarity between objects for re-identification, allowing continued tracking of objects that have exited and potentially re-entered the scene, while maintaining tolerance for temporary occlusions and camera view changes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the re-identification threshold is set to a lower value to allow re-identification of similar objects, then the tracking continuity is improved, but the risk of misidentifying different objects as the same object increases

Engineering Contradiction:
Improvetracking continuityVSAvoidobject identification accuracy
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The re-identification threshold is made dynamic rather than static. The system adjusts the threshold based on whether an object has exited the scene: using a first (lower) threshold for objects within the scene to maintain tracking continuity, and a second (higher) threshold for re-identifying objects that have re-entered to prevent misidentification. This dynamic adjustment resolves the contradiction by adapting the threshold to the specific tracking context.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes the re-identification threshold parameter based on the tracking state. When an object exits and re-enters the scene, the threshold is adjusted from a first value to a second value, which is configured to prevent misidentification. This parameter change allows the system to maintain tracking continuity for occluded objects while preventing false re-identification of similar but different objects.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If the re-identification threshold is increased to prevent misidentification, then the object identification accuracy is improved, but the tracking continuity for occluded objects deteriorates

Engineering Contradiction:
Improveobject identification accuracyVSAvoidtracking continuity
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system dynamically switches between two threshold values based on the tracking scenario. A first threshold maintains tracking continuity during occlusions, while a second threshold prevents misidentification after scene exit. This dynamic approach ensures that high accuracy is achieved when needed without sacrificing tracking continuity during normal occlusions.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The re-identification threshold parameter is changed based on the object's tracking history. When an object has exited the scene, the threshold is set to a higher value to prevent misidentification. This parameter change ensures that identification accuracy is maintained for re-entering objects while allowing more flexible matching for continuously tracked objects.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If tracking is terminated when an object exits the scene to save computational resources, then the computational efficiency is improved, but the ability to track re-entering objects deteriorates

Engineering Contradiction:
Improvecomputational efficiencyVSAvoidre-entry detection capability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system performs preliminary action by adjusting the re-identification threshold in advance before an object re-enters the scene. When an object exits, the threshold is adjusted to a second value that prevents misidentification. This preliminary adjustment ensures that when the object re-enters, the system can correctly identify it as a new object rather than incorrectly matching it with the previous object, thus maintaining reliable re-entry detection.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses feedback from the object exit event to adjust the re-identification threshold. When an object exits the scene, this feedback triggers a threshold adjustment that affects subsequent re-identification attempts. This feedback mechanism ensures that computational resources are used efficiently while maintaining the ability to correctly detect and track re-entering objects.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4564314B1Method of tracking objects
Publication Date: 2025.11.19 AXIS
  • EP4564314B1 patent drawingFigure 1
  • EP4564314B1 patent drawingFigure 2
  • EP4564314B1 patent drawingFigure 3

AI summary

According to an aspect, there is provided a computer-implemented method of tracking objects in a video sequence of a scene, the method comprising: determining a location of a sink in the scene where objects exit the scene and a location of a source where objects enter the scene; tracking a first object moving in the scene using a re-identification algorithm, wherein the first object is associated with a re-identification threshold of the re-identification algorithm; detecting that the first object has exited the scene at the sink; and responsive to detecting that the first object has exited the scene at the sink, adjusting the re-identification threshold associated with the first object such that a probability that the re-identification algorithm re-identifies a second object, entering the scene at the source after the first object has exited the scene at the sink, as the first object is reduced.