Computer Vision Annotator

Odixcity Consulting; South Africa; Remote; ZAR 699.000 - 1.065.000; Vollzeit

Zusammenfassung

Odixcity Consulting is seeking a Computer Vision Annotator to label image and video data for training ML models. This remote position supports projects worldwide and requires meticulous annotation to help improve object detection, segmentation, and tracking.

You will use tools like CVAT, Label Studio, and 3D Slicer to draw boxes, polygons, key points, and 3D cuboids, and you will help refine guidelines with data scientists and managers.

Qualifikationen

• 2+ years of experience in data annotation for computer vision or ML projects.

Aufgaben

• Perform high-precision annotation on visual data using specialized tools.
• Drawing accurate bounding boxes, polygons, and polylines around objects of interest.
• Placing key points for pose estimation or landmark detection.
• Creating 3D cuboids and annotating point cloud data for spatial awareness applications.
• Conducting semantic/instance segmentation to label every pixel in an image.
• Interpreting complex labeling taxonomies and refining annotation guidelines with project managers and data scientists.
• Using annotation tools features and providing feedback to improve tool functionality and workflow efficiency.
• Identifying challenging data patterns, edge cases, and potential biases to improve model robustness.
• Contributing to annotation best practices, workflow optimizations, and training materials for new annotators.
• Assisting in basic data preparation tasks such as resizing images and converting data formats.
• Tracking KPIs such as throughput, precision, and recall; participating in calibration sessions.

Kenntnisse

Data annotation

Quality assurance

English communication

Tools

CVAT

Label Studio

3D Slicer

Supervisely

Jobbeschreibung

Job Title

Computer Vision Annotator

Location

Remote (Worldwide)

Job Summary

The Computer Vision Annotator is responsible for accurately labeling and annotating image and video data to support the development and training of computer vision and machine learning models. This role plays a critical part in improving model performance by creating high-quality datasets used for object detection, image classification, segmentation, tracking, and other visual recognition tasks.

Responsibilities

• Perform high-precision annotation on visual data using specialized tools. Tasks include:
• Drawing accurate bounding boxes, polygons, and polylines around objects of interest.
• Placing key points for pose estimation or landmark detection.
• Creating 3D cuboids and annotating point cloud data for spatial awareness applications (e.g., autonomous vehicles, robotics).
• Conducting semantic/instance segmentation to label every pixel in an image.
• Interpret complex labeling taxonomies and project specifications. Identify ambiguous cases and collaborate with project managers and data scientists to refine annotation guidelines for clarity and scalability.
• Utilize advanced features of annotation tools such as CVAT, Label Studio, 3D Slicer, or Amazon SageMaker Ground Truth. Provide feedback to engineering teams to improve tool functionality and workflow efficiency.
• Proactively identify challenging data patterns, edge cases, and potential biases in datasets. Escalate these findings to help improve model robustness and mitigate bias.
• Contribute to the development of annotation best practices, workflow optimizations, and training materials for new annotators.
• Assist in basic data preparation tasks such as resizing images, converting data formats, or ensuring sensor data synchronization (e.g., time-stamp alignment for video and LiDAR).
• Assist in tracking key performance insights (KPIs) such as annotation throughput, precision, and recall. Participate in calibration sessions to ensure consistency across the team.

Requirements

• Minimum of 2 years of experience in data annotation, data labeling, or quality assurance specifically for computer vision or Machine Learning projects.
• Demonstrated proficiency with a variety of annotation techniques including bounding boxes, polygons, key points, 3D cuboids, and segmentation masks.
• Extensive hands-on experience with industry-standard annotation tools like CVAT, label Studio, 3D Slicer, Supervisely.
• Solid understanding of computer vision concepts and terminology (e.g., object detection, image classification, LiDAR, point clouds).
• Familiarity with quality control processes such as consensus labeling, audit sampling, gold-task seeding, and inter-annotator agreement metrics.
• Exceptional attention to detail with an unwavering commitment to accuracy and consistency. Proven ability to maintain high-quality output even when labeling repetitive tasks.
• Strong written and verbal communication skills in English. Ability to document workflows clearly and collaborate effectively teams and clients.