Object Detection False Positive Filtering

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image recognition technologies face challenges in effectively protecting privacy by accurately detecting and obscuring human faces and license plates in images, as high-recall settings often result in increased false positives, compromising privacy.

Innovation Solution

A system comprising an object detector module, a false positive detector module, and an image blurring module processes images to detect faces and license plates with high recall, uses context information to filter out false positives, and applies blurring algorithms to ensure privacy protection, employing a combination of sliding window algorithms, feature detectors, and neural networks to differentiate between actual and false positives.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If face detection algorithms are tuned for high recall, then the likelihood of detecting every face is increased, but the number of false positives increases

Engineering Contradiction:

Improveface detection recallVSAvoidfalse positive rate

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent segments the face detection process into multiple independent stages: initial high-recall detection, candidate region generation, and verification filtering. Each stage operates with different sensitivity thresholds, allowing the system to maintain high overall recall while reducing false positives through progressive filtering.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces candidate region proposals as an intermediary between initial detection and final verification. These candidate regions act as a buffer that filters out obvious false positives before they reach the verification stage, while preserving regions that might contain actual faces.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If multiple detection layers and context information processing are added, then false positives are reduced, but device complexity increases

Engineering Contradiction:

Improvefalse positive filtering accuracyVSAvoiddetection system structure

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary filtering of candidate regions before full verification. By pre-processing candidate regions and eliminating obvious false positives early in the pipeline, the system reduces the computational burden on subsequent verification stages while maintaining high precision.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent applies different levels of verification to different candidate regions. High-confidence candidates receive full verification while low-confidence candidates are either quickly rejected or subjected to more intensive analysis only if necessary, optimizing the balance between precision and computational complexity.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS9104914B1Object detection with false positive filtering

Publication Date: 2015.08.11 GOOGLE LLC

AI summary

Embodiments of this invention relate to detecting and blurring images. In an embodiment, a system detects objects in a photographic image. The system includes an object detector module configured to detect regions of the photographic image that include objects of a particular type at least based on the content of the photographic image. The system further includes a false positive detector module configured to determine whether each region detected by the object detector module includes an object of the particular type at least based on information about the context in which the photographic image was taken.