Object Detection Device for Efficient Teacher Data Creation

Solution Overview

Problem

Creating teacher data for object detection devices is labor-intensive and time-consuming, as users must manually select and classify numerous detected objects from images, which is inefficient and laborious.

Innovation Solution

An object detection device that includes a detecting unit for identifying objects in images, an accepting unit for displaying a graph of detected objects and allowing user input to select and classify them, and a learning unit for updating the dictionary based on the user-generated data, enabling efficient creation of teacher data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual selection and classification of detected objects is performed to create teacher data, then high-quality teacher data can be generated, but the process becomes labor-intensive and time-consuming

Engineering Contradiction:

Improvequality of teacher dataVSAvoidtime required for creating teacher data

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system automatically performs object detection and preliminary classification using the dictionary matching unit, allowing the creation of teacher data with minimal manual intervention. The automatic classification based on dictionary matching reduces the need for manual labor while maintaining data quality.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces manual mechanical classification with automated dictionary matching and feature comparison. The system automatically compares detected object features with dictionary entries to classify objects, substituting human manual work with automated computational processes.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If manual selection and classification of detected objects is performed to create teacher data, then high-quality teacher data can be generated, but the process becomes labor-intensive

Engineering Contradiction:

Improvequality of teacher dataVSAvoidoperational simplicity

Core Design Contradiction:

ReliabilityVSEase of operation

Solution Approach 1:

The system automatically performs object detection and preliminary classification using the dictionary matching unit, allowing the creation of teacher data with minimal manual intervention. The automatic classification based on dictionary matching reduces the need for manual labor while maintaining data quality.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces manual mechanical classification with automated dictionary matching and feature comparison. The system automatically compares detected object features with dictionary entries to classify objects, substituting human manual work with automated computational processes.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Reliability

If all detected images are displayed for user selection, then complete control over teacher data creation is achieved, but the complexity of the process increases

Engineering Contradiction:

Improvecontrol over teacher dataVSAvoidprocess complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The system extracts and displays only relevant detected objects that match dictionary entries, filtering out unnecessary information. This selective display approach maintains user control while reducing the complexity of the interface and process by focusing only on significant detections.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system performs preliminary automatic classification and filtering of detected objects before presenting them to the user. By pre-processing the detection results and organizing them according to dictionary categories, the system reduces the complexity of subsequent user interactions.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20240355113A1Object detection device, object detection method, and recording medium

Publication Date: 2024.10.24 NEC CORP

AI summary

Provided is an object detection device for efficiently and simply selecting an image for creating instructor data on the basis of the number of detected objects. The object detection device is provided with: a detection unit for detecting an object from each of a plurality of input images using a dictionary; an acceptance unit for displaying, on a display device, a graph indicating the relationship between the input images and the number of subregions in which the objects are detected, and displaying, on the display device, in order to create instructor data, one input image among the plurality of input images in accordance with a position on the graph accepted by operation of an input device; a generation unit for generating the instructor data from the input image; and a learning unit for learning a dictionary from the instructor data.