Collaborative Image Label Updating for Low-Confidence Object Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing object identification techniques for safe vehicle travel, such as semantic segmentation, struggle with identifying objects not registered in the dictionary and lack effective handling of low identification confidence objects.

Innovation Solution

An information processing apparatus and method that includes an image analysis unit for object identification, a low-confidence region extraction unit, and a label updating unit, which uses communication information to update labels of low-confidence regions based on inter-vehicle communication, ensuring reliable object identification and safe travel.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If semantic segmentation is used for object identification, then identification speed is improved, but identification accuracy deteriorates for objects not registered in the dictionary

Engineering Contradiction:

Improveidentification speedVSAvoididentification accuracy

Core Design Contradiction:

SpeedVSMeasurement precision

Solution Approach 1:

The patent merges multiple identification results from different mobile devices through a result integration unit. By combining detection data from multiple sources, the system achieves both fast identification (inherited from semantic segmentation) and high accuracy (through multi-source verification), resolving the contradiction between speed and precision for unregistered objects

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The system implements feedback mechanisms where identification results are continuously refined through inter-device communication. Low-confidence detections trigger additional verification rounds, allowing the system to maintain high-speed initial detection while improving accuracy through iterative feedback from multiple devices

Inventive Principle:

Principle #23Feedback

2Device complexity

If only camera-based image analysis is used, then device complexity is reduced, but identification reliability deteriorates for low-confidence objects

Engineering Contradiction:

Improvedevice complexityVSAvoididentification reliability

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The patent makes mobile devices universal by enabling them to perform both independent identification and collaborative verification functions. Each device maintains its simple camera-based analysis capability while also serving as part of a distributed identification network, achieving high reliability without increasing individual device complexity

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The communication unit acts as an intermediary that enables reliability improvement without adding complex hardware. By mediating information exchange between devices, the system achieves enhanced identification reliability for low-confidence objects while keeping each individual device relatively simple

Inventive Principle:

Principle #24Intermediary (Mediator)

3Ease of operation

If dictionary-based semantic segmentation is used, then ease of operation is improved, but adaptability deteriorates for unknown objects

Engineering Contradiction:

Improveease of operationVSAvoidadaptability to unknown objects

Core Design Contradiction:

Ease of operationVSAdaptability or versatility

Solution Approach 1:

The patent segments the identification process into confidence-based categories: high-confidence objects use fast dictionary matching (easy operation), while low-confidence objects trigger collaborative verification (adaptability). This segmentation allows the system to maintain ease of operation for common objects while adapting to unknown objects through additional processing layers

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system dynamically adjusts its operation mode based on identification confidence. For high-confidence detections, it operates in simple dictionary-matching mode (easy operation). For low-confidence or unknown objects, it dynamically switches to collaborative verification mode (adaptability), resolving the contradiction between operational simplicity and adaptability

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS12190595B2Information processing apparatus, information processing system, and information processing method

Publication Date: 2025.01.07 SONY GROUP CORP

AI summary

Provided is an image analysis unit which analyzes a captured image of a camera mounted on a mobile device, executes object identification of an image, and sets a label as an identification result to an image region; a low-confidence region extraction unit which extracts a region with low confidence of object identification from an image analysis result; and a label updating unit which updates a label of the low-confidence region on the basis of information received via a communication unit. The label updating unit updates a label in a case where a matching rate between an object region analyzed from information received via the communication unit and the low-confidence region is equal to or greater than a specified threshold.