Image Annotation Quality Control Specialist

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. We connect people to hands-on projects that shape how modern AI systems behave and grow careers in a fast-moving field.

Creating an OpenTrain account is free. We support flexible, remote work that fits around life and other commitments while contributors learn skills that matter across AI teams.

Why this AI training work matters

Human review and quality control are essential to reliable computer vision: trained models depend on accurate labels and consistent annotations. Your corrections will directly improve model performance for industrial visual inspection in refinery environments.

This role is part of the human-in-the-loop process that turns raw footage into trustworthy training data—work that is remote, flexible, and at the core of how real-world AI is built.

Role overview

We are hiring an Image Annotation Quality Control Specialist (contract, part-time) to review and correct auto-generated annotations in an 8,000-frame dataset captured from refinery footage. The dataset contains mostly auto-annotations; some frames will require corrections or added labels.

This intermediate-level role requires about 20+ hours per week, is 100% remote (worldwide), uses CVAT for annotation work, and pays a fixed price of USD 1,200 for the contract.

What you'll do

Perform visual QA across the dataset, focusing on accuracy and consistency of both classification tags and bounding-box labels.

Correct auto-annotations where the model missed or mislabeled objects, and add missing annotations on frames where auto-labels are incomplete.

• Review frames in CVAT and apply corrections to classification labels and bounding boxes.
• Ensure annotations follow existing guidelines and maintain consistency across similar clear frames.
• Flag ambiguous or poor-quality frames for team review when necessary.

Requirements

You must be an Intermediate-level annotator comfortable working independently on a contractor basis. The role requires strong attention to detail and prior experience reviewing and correcting auto-annotations.

Fluent English proficiency is required for clear communication and documentation.

• Experience with image annotation QC and correcting auto-generated labels.
• Familiarity with bounding-box workflows and classification labeling.
• Comfort using CVAT or similar annotation tools.

Project details, schedule & pay

Dataset size: 8,000 image frames derived from refinery footage. The model performs consistently on clear frames but will need human corrections on edge cases, occlusions, or missing labels.

Time commitment: 20+ hours per week. Employment type: Contractor, part-time. Payment: Fixed-price contract — USD 1,200 total for the project.

• Label types: Classification and bounding box.
• Tool: CVAT will be used for all annotation and QC tasks.
• Location: Remote — open to contributors worldwide.

How to apply and sample data

To apply, create a free OpenTrain account, complete your profile, and submit your application for this listing. We evaluate candidates based on relevant annotation/QC experience and attention to detail.

A small set of sample frames is available for review. Use the link below to inspect dataset examples before applying (you may reference specific concerns or questions in your application).

• Sample frames: https://mega.nz/folder/wZQjHDjR#8EQ7FCugT_o3oj1aNtiarQ
• Include brief notes about your experience correcting auto-annotations and any familiarity with industrial or inspection imagery.

Who should apply

Apply if you have hands-on experience QAing image annotations, are comfortable with CVAT, and want flexible, remote contract work that directly improves industrial computer vision models.

This role suits annotators who pay close attention to visual detail, enjoy systematic correction work, and can commit 20+ hours weekly.