Manual Structural Crack Segmentation Annotator

About OpenTrain

OpenTrain AI is the hiring and contracting organization for this role. OpenTrain helps people find and build careers in AI training and data labeling, connecting contributors with projects where careful human work helps improve artificial intelligence.

This is a worldwide, remote contractor opportunity with a fixed project payment of $170. Creating an OpenTrain account is free, and candidates can apply in minutes.

• Contractor and part-time project
• Worldwide eligibility
• Expected commitment of at least 20 hours per week
• Expected duration of under one month
• Fixed project payment: $170

About AI Training and Image Annotation

AI training depends on people who prepare and review high-quality examples. In image annotation, contributors identify visual features and outline them precisely so AI systems can learn how objects, defects, and other structures appear in real-world imagery.

This project focuses on careful human judgment rather than automation. Your independent annotations will support ground-truth development and an inter-annotator agreement study for structural crack assessment.

• Work on a real AI training and data-labeling project
• Use human visual judgment to distinguish structural cracks from similar patterns
• Help create reliable reference data for research and model evaluation

The Role

As a Manual Structural Crack Segmentation Annotator, you will create pixel-level polygon or brush annotations of visible structural cracks in UAV bridge imagery. A representative portion of the dataset includes 12 high-resolution images measuring 8192 × 5460 pixels.

The images may contain thin or branching cracks, complex concrete textures, joints, shadows, stains, vegetation, and distracting background patterns. You will need to zoom carefully, work independently, and make consistent decisions across challenging visual conditions.

A smaller pilot will be completed first. The remaining dataset will proceed if your annotation quality and consistency meet the project requirements.

• Experience level: Intermediate
• Data type: high-resolution images
• Annotation type: manual polygon and pixel-level segmentation
• Software: Roboflow
• Subject area: civil engineering, structural engineering, bridge inspection, and infrastructure condition assessment

What You’ll Do

Annotate only visible structural cracks located on the defined structural surfaces of the bridge. Work entirely independently from existing labels and model predictions, using Roboflow's manual polygon or brush tools and standard navigation features.

When a feature cannot confidently be classified as a structural crack after reasonable inspection and zooming, do not force a decision. Mark the image or region as uncertain or add a review comment, and be prepared to explain your screening decisions clearly.

• Manually segment thin, branching, and difficult-to-see structural cracks
• Zoom and inspect high-resolution UAV bridge images carefully
• Exclude construction joints, expansion joints, formwork lines, and structural edges
• Exclude stains, shadows, vegetation, cables, soil fissures, and ground cracks
• Exclude background construction elements and other crack-like patterns outside the target bridge structure
• Complete a pilot before continuing with the remaining dataset if quality standards are met
• Add uncertainty markings or review comments when classification is not confident

Requirements

This project is best suited to candidates with relevant experience in manual polygon or segmentation annotation. Experience in civil engineering, infrastructure or defect inspection, bridge inspection, structural assessment, or infrastructure condition assessment is strongly preferred.

You should be able to discuss your relevant experience with structural crack segmentation, maintain strong accuracy over detailed imagery, and explain screening decisions clearly. Basic English proficiency is required.

• Intermediate-level experience
• At least 20 hours of weekly availability
• Strong accuracy and attention to detail
• Relevant manual segmentation or polygon annotation experience
• Preferred civil, infrastructure, bridge, structural, or defect-inspection experience
• Ability to discuss relevant structural crack segmentation experience
• Ability to explain screening decisions clearly
• Basic English proficiency

Strict Manual Annotation Standards

All annotations must be produced manually and independently. The project does not permit AI-assisted annotation, automatic segmentation, model-generated pre-labeling, or any similar automation because the research requires human-generated reference annotations.

• Do not use SAM, SAM 2, SAM 3, or other AI-assisted tools
• Do not use Smart Polygon, Label Assist, Auto Label, or Box Prompting
• Do not use model-generated predictions or existing labels as annotation guidance
• Use only manual polygon or brush tools and standard navigation in Roboflow

How to Apply Through OpenTrain

Create a free OpenTrain account and apply in minutes. Be ready to describe your experience with manual segmentation, structural cracks, bridge or infrastructure inspection, and careful visual screening.

If selected, you will begin with the pilot assignment. Continued work on the remaining dataset depends on meeting the project's required annotation quality and consistency.

• Apply through OpenTrain AI
• Highlight relevant civil, structural, bridge, or infrastructure experience
• Describe your manual polygon or segmentation background
• Confirm your availability of at least 20 hours per week
• Be prepared to discuss how you handle uncertain visual features