Adaptive Video Object Detection Thresholds

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Setting detection thresholds in video object detection is challenging due to varying detection conditions across different scene areas, leading to issues with false positives and false negatives, which existing methods fail to address efficiently.

Innovation Solution

A method and system that adapt object detection thresholds by segmenting a scene into areas with differing detection conditions, tracking objects as they move between these areas, and adjusting thresholds based on confidence values to ensure accurate detection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Object-affected harmful factors

If a high detection threshold is set to reduce false positives, then false positives are reduced, but false negatives increase

Engineering Contradiction:

Improvefalse positivesVSAvoidfalse negatives

Core Design Contradiction:

Object-affected harmful factorsVSReliability

Solution Approach 1:

The scene is divided into multiple scene areas with different detection conditions. Each scene area has its own optimized detection threshold, allowing high thresholds in areas with good detection conditions (reducing false positives) and low thresholds in areas with poor detection conditions (reducing false negatives).

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Different detection thresholds are applied to different spatial locations within the scene based on local detection conditions. This allows each region to have a threshold optimized for its specific characteristics, rather than using a single global threshold that compromises overall performance.

Inventive Principle:

Principle #3Local quality

2Reliability

If a low detection threshold is set to reduce false negatives, then false negatives are reduced, but false positives increase

Engineering Contradiction:

Improvefalse negativesVSAvoidfalse positives

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The scene is divided into multiple scene areas with different detection conditions. Each scene area has its own optimized detection threshold, allowing low thresholds in areas with poor detection conditions (reducing false negatives) and high thresholds in areas with good detection conditions (reducing false positives).

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Inventive Principle:

Principle #3Local quality

3Device complexity

If a single detection threshold is used for the entire scene, then device complexity is reduced, but detection accuracy worsens due to varying conditions

Engineering Contradiction:

Improvethreshold configurationVSAvoiddetection accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The scene is automatically divided into multiple scene areas based on detection conditions, with each area receiving a customized threshold. This maintains simplicity in the overall system while improving local detection accuracy through targeted threshold optimization.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system automatically determines detection conditions for different scene areas and sets appropriate thresholds without requiring manual configuration. This self-calibration process improves detection accuracy while keeping the user interface simple.

Inventive Principle:

Principle #25Self-service

Data Source

PatentEP4600913B1Method and system for calibration of detection thresholds in video object detection

Publication Date: 2026.04.01 AXIS

Figure 1
Figure 2a~2b
Figure 3a~3b

AI summary

A method for detecting objects in a scene (2) captured by a camera is disclosed. The method comprises capturing video of the scene, segmenting a first image in into at least a first and second scene area (A1, A2), the first and second scene areas having properties such that object detection conditions are expected to differ between the first and second scene areas. A first object detection threshold is set for the first scene area. A first object (3) in the first scene area (A1) in the first image is detected, a first confidence value for the detection of the first object being above the first object detection threshold. In a plurality of images subsequent to the first image, the first object is tracked (4) as it moves into the second scene area (A2). A second confidence value for a detection of the first object in the second scene area is determined. A second object detection threshold for the second scene area is set such that the second confidence value for the detection of the first object is above the second object detection threshold. In subsequent images, the first object detection threshold is used for detecting objects in the first scene area and the second object detection threshold is used for detecting objects in the second scene area.