Self-optimized Object Detection via Dynamic Detector Selection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing surveillance video systems face challenges in reliably detecting objects continuously under varying conditions such as weather and time of day due to significant object appearance variations caused by lighting effects, object pose changes, and multiple environmental factors, leading to inefficiencies in manual monitoring and high computational costs.

Innovation Solution

The system dynamically selects and applies a set of efficient detector models that cover multiple appearance subspaces by classifying detectors as foreground or background based on calibrated confidence scores, ensuring that suitable detectors are run frequently or sporadically depending on environmental conditions, thereby adapting to changing scenarios without extensive retraining.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If a single detector is used for object detection, then the system is simple to operate, but the detection reliability deteriorates under varying environmental conditions

Engineering Contradiction:

Improvedetector application simplicityVSAvoiddetection reliability

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent divides the detector pool into foreground detectors (for active object detection) and background detectors (for environmental adaptation), allowing the system to switch between different detector subsets based on environmental conditions. This segmentation enables reliable detection across varying conditions while maintaining operational simplicity through automated selection.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system dynamically selects and switches between different detectors based on real-time environmental conditions and detector performance metrics. This dynamic adaptation allows the system to maintain high detection reliability under varying conditions without requiring manual reconfiguration, resolving the contradiction between simplicity and reliability.

Inventive Principle:

Principle #15Dynamics

2Reliability

If multiple detectors are applied to cover different environmental conditions, then the detection reliability improves, but the computational cost increases

Engineering Contradiction:

Improvedetection reliabilityVSAvoidcomputational load

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent applies only the necessary subset of detectors (foreground or background) based on current environmental conditions rather than running all detectors continuously. This partial action approach maintains detection reliability by selecting appropriate detectors while reducing computational load by excluding unnecessary ones from each processing cycle.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The system changes operational parameters by switching between different detector configurations (foreground/background modes) based on environmental conditions. This parameter change strategy allows the system to maintain reliability across conditions while optimizing computational efficiency by adapting the detector application rate and selection criteria to current needs.

Inventive Principle:

Principle #35Parameter changes

3Adaptability or versatility

If detectors are frequently switched to adapt to environmental changes, then the adaptability improves, but the system stability deteriorates due to drifting issues

Engineering Contradiction:

Improveenvironmental adaptabilityVSAvoidsystem stability

Core Design Contradiction:

Adaptability or versatilityVSStability of the object's composition

Solution Approach 1:

The patent implements periodic evaluation and switching between foreground and background detectors based on environmental condition thresholds and detector performance metrics. This periodic action provides structured adaptability to environmental changes while maintaining system stability through regular, controlled transitions rather than continuous switching, reducing drifting issues.

Inventive Principle:

Principle #19Periodic action

Solution Approach 2:

The system uses feedback from detector performance metrics and environmental condition monitoring to control detector selection and switching. This feedback mechanism enables adaptive response to environmental changes while maintaining stability by using performance-based criteria to determine when switching is appropriate, preventing unnecessary transitions that could cause drifting.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS10089551B2Self-optimized object detection using online detector selection

Publication Date: 2018.10.02 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

Embodiments are directed to an object detection system having at least one processor circuit configured to receive a series of image regions and apply to each image region in the series a detector, which is configured to determine a presence of a predetermined object in the image region. The object detection system performs a method of selecting and applying the detector from among a plurality of foreground detectors and a plurality of background detectors in a repeated pattern that includes sequentially selecting a selected one of the plurality of foreground detectors; sequentially applying the selected one of the plurality of foreground detectors to one of the series of image regions until all of the plurality of foreground detectors have been applied; selecting a selected one of the plurality of background detectors; and applying the selected one of the plurality of background detectors to one of the series of image regions.