Automated Defect Analysis in Sewer Infrastructure Using Computer Vision

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for evaluating defects in sewer and water infrastructure are inconsistent due to human error and operator variability, despite advancements in computer image recognition and cloud computing, leading to inconsistent defect assessment and reporting.

Innovation Solution

A software tool that analyzes video footage of sewer systems using a trained weight system to determine the severity and type of defects, providing a color-coded report free from human error, utilizing object detection algorithms like YOLOv3 in Tensorflow to identify and categorize defects accurately.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If human operators manually evaluate defects in sewer systems, then flexibility in assessment is maintained, but consistency and objectivity deteriorate due to human error and operator variability

Engineering Contradiction:

Improveconsistency of defect assessmentVSAvoidcomplexity of evaluation system

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces manual visual inspection by human operators with an automated image recognition system. The system captures images of sewer defects using cameras and processes them through computer vision algorithms to automatically identify, classify, and assess defect severity, eliminating human error and operator variability while maintaining assessment flexibility through programmable detection criteria.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system creates digital copies of physical defect images and analyzes these copies through computational models. By working with image data rather than directly manipulating physical infrastructure, the system enables repeated, consistent analysis of the same defect from multiple angles and by different operators, ensuring reliable and reproducible assessment results.

Inventive Principle:

Principle #26Copying

2Measurement precision

If automated image recognition systems are used to analyze defects, then objectivity and consistency improve, but measurement precision deteriorates due to inability to capture subtle defect characteristics

Engineering Contradiction:

Improveaccuracy of defect detectionVSAvoidobjectivity of defect assessment

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The system divides the sewer infrastructure into multiple inspection zones and captures images at different locations, angles, and distances. By segmenting the inspection process into multiple controlled observations, the system builds comprehensive defect profiles that capture subtle characteristics while maintaining objective, repeatable measurements through standardized image acquisition protocols.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system varies multiple parameters during image capture including lighting conditions, camera angle, focal length, and resolution settings to optimize defect visibility. By changing these parameters systematically, the system enhances the detection of subtle defect features while maintaining consistent, objective assessment criteria through programmable control of inspection conditions.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11967056B2Systems, methods and apparatuses for detecting and analyzing defects in underground infrastructure systems

Publication Date: 2024.04.23 INFRASTRUCTURE DL LLC

AI summary

Systems, methods and devices for performing analysis of surfaces, external and internal, including public works. The invention employs analytical software that provides results independent of the limitations of the operator.