False-Report Elimination in Intelligent Security Cameras

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current intelligent security cameras face high computing costs and privacy concerns due to cloud computing, while edge computing provides an unsatisfactory experience with frequent false positives, necessitating a solution to reduce costs, improve stability, and enhance user experience.

Innovation Solution

A false positive elimination method that acquires and analyzes images to determine encryption features, compares historical and current image information, and verifies target objects using a split network architecture with deep learning and knowledge distillation, reducing device and cloud computing loads while protecting user privacy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If cloud computing is used for intelligent security camera analysis, then intelligent experience and detection accuracy are improved, but computing costs and server pressure increase

Engineering Contradiction:

Improvedetection accuracyVSAvoidcomputing cost

Core Design Contradiction:

Measurement precisionVSLoss of energy

Solution Approach 1:

The system segments the computing tasks between edge device and cloud server. The edge device performs local object detection and generates initial detection results, while the cloud server performs verification only on uncertain cases. This segmentation reduces cloud computing pressure and costs while maintaining high detection accuracy through collaborative processing.

Inventive Principle:

Principle #1Segmentation

2Productivity

If cloud computing is used for image analysis, then intelligent experience is improved, but privacy security deteriorates due to image upload requirements

Engineering Contradiction:

Improveintelligent experienceVSAvoidprivacy leakage risk

Core Design Contradiction:

ProductivityVSObject-affected harmful factors

Solution Approach 1:

The system introduces an intermediary verification mechanism where the cloud server verifies only encrypted feature information and detection results, not the original images. This intermediary approach allows cloud-based intelligence while protecting privacy by preventing direct access to image data.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Loss of energy

If edge computing is used for local processing, then computing costs and privacy protection are improved, but detection accuracy and intelligent experience deteriorate due to false positives

Engineering Contradiction:

Improvecomputing costVSAvoiddetection accuracy

Core Design Contradiction:

Loss of energyVSMeasurement precision

Solution Approach 1:

The system implements a feedback mechanism where the cloud server verifies detection results from the edge device and provides correction feedback. The edge device sends detection results to the cloud, which verifies them using a different model, and returns verification results to improve future local detections, reducing false positives while maintaining low computing costs.

Inventive Principle:

Principle #23Feedback

4Measurement precision

If cloud server verifies all detection results, then detection accuracy is improved, but server pressure and computing costs increase during peak periods

Engineering Contradiction:

Improvedetection accuracyVSAvoidserver pressure

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system applies partial verification action where the cloud server verifies only a subset of detection results - specifically those from uncertain or low-confidence detections. High-confidence detections are accepted without verification, reducing server pressure while maintaining overall detection accuracy through selective verification.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20240330507A1Method, Device, Equipment and Medium for False-Report Elimination

Publication Date: 2024.10.03 HANGZHOU MEARI TECH CO LTD

AI summary

A false positive elimination method includes: analyzing a target object to obtain an encryption feature and current position information of the target object; acquiring current image information of the target object, and comparing local historical image information with the current image information to obtain a comparison result; detecting the current image information according to the current position information when the historical image information is inconsistent with the current image information, and sending the encryption feature to a cloud when the target object is detectable in the current image information, so that the cloud determines actual position information of the target object based on the encryption feature; and acquiring the actual position information, verifying the target object based on the actual position information, determining the target object as a false positive object in response to the passed verification, and performing a false positive elimination operation on the false positive object.