Hazard Prediction Using Image Recognition for Construction Safety

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Hazard prediction activities at construction sites are inconsistent due to varying worker experience, personality, and physical condition, leading to potential accidents despite efforts to prevent them.

Innovation Solution

A method and apparatus that utilize image recognition to match site conditions with past accident data, providing workers with accurate hazard information and safety measures, and controlling tool access based on correct safety measures implementation, ensuring safety regardless of worker skills or personalities.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If hazard prediction activities are performed by workers based on their experience and judgment, then the prediction process is simple and quick, but the accuracy and consistency of hazard prediction vary due to differences in worker experience, personality, and physical condition

Engineering Contradiction:

Improvespeed of hazard predictionVSAvoidconsistency of hazard prediction

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent replaces the mechanical human judgment system with an automated image recognition system. The system captures images of the work site, automatically identifies hazards through image processing, and provides consistent hazard predictions without relying on worker experience or physical condition, thereby resolving the contradiction between prediction speed and consistency.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The hazard prediction system performs self-service by automatically analyzing work site images and generating hazard predictions without requiring worker intervention. The system independently processes images, identifies hazards, and provides predictions, eliminating variability caused by different workers' experiences and physical states while maintaining consistent accuracy.

Inventive Principle:

Principle #25Self-service

2Adaptability or versatility

If workers with low experience perform hazard prediction activities, then labor costs are reduced and more workers can participate, but accident risk increases due to insufficient knowledge

Engineering Contradiction:

Improveparticipation of workers with varying experience levelsVSAvoidaccident risk

Core Design Contradiction:

Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

The system replaces worker-based hazard prediction with an automated image recognition system that objectively identifies hazards. This substitution eliminates the accident risk associated with inexperienced workers while maintaining broad participation, as any worker can use the system regardless of their experience level.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces an intermediary system (image recognition system) between the worker and the hazard prediction process. This intermediary automatically analyzes work site conditions and provides accurate hazard predictions, serving as a mediator that bridges the gap between worker participation and safe prediction practices, thereby reducing accident risk while maintaining accessibility.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Device complexity

If workers perform hazard prediction based on their personal judgment, then the process requires minimal equipment, but prediction accuracy varies due to personality differences and physical condition

Engineering Contradiction:

Improvesimplicity of prediction processVSAvoidaccuracy of hazard prediction

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent replaces the simple but variable human judgment process with an automated image recognition system. The system uses cameras and image processing algorithms to objectively measure and identify hazards, providing consistent accurate predictions without the influence of worker personality or physical condition, while the overall system remains relatively simple to deploy.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11967223B2Danger prediction method and danger prediction device

Publication Date: 2024.04.23 NIPPON TELEGRAPH & TELEPHONE CORP

AI summary

The purpose of the present invention is to provide a hazard prediction method that can perform hazard prediction activities so as to prevent accidents, regardless of skills and personalities of workers.A hazard prediction method according to the present invention includes: photographing an outdoor facility along with environment to capture an image; collating the image with accident data including photos of outdoor facilities along with environment and accidents occurred at open-air facilities in the photos to detect, when a photo similar to the image is found, an accident occurred at an open-air facility in the photo from the accident data; extracting, from a safety measure list describing accidents, hazardous spots related to the accidents, and safety measures for prevention of the accidents, hazard information on a hazardous spot and a safety measure corresponding to the detected accident; receiving input information that is a safety measure devised by a worker for the hazardous spot; and comparing the input information with the hazard information to check that the input information and the hazard information match each other.