Image Labeler — Bounding Boxes for Purple-Outlined Game Characters

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. We connect contributors with real projects that teach AI systems how to see, hear, and reason, and creating an OpenTrain account is free.

This role is posted on OpenTrain because the project needs reliable image labelers who understand object-detection workflows and can produce YOLO-ready bounding boxes.

Why AI Training Work Matters

Human labeling is the foundation of modern AI — people create the examples that models learn from. This work is 100% remote, often part-time and flexible, and a practical way to gain experience on cutting-edge AI projects.

Labeling for object detection directly shapes how computer vision models recognize objects in games, apps, and research.

The Role

You will label characters that are outlined in purple in a set of game images by drawing accurate axis-aligned bounding boxes so they can be used to train a YOLO object-detection model.

Images typically contain 1–5 characters. Examples and a short labeling guide will be provided with each task.

• Label type: Bounding box (object detection).
• Annotation tool: Roboflow.
• Deliverable format: YOLO-compatible labels.

What You'll Do Day to Day

Work through batches of images in Roboflow and mark every character that has a purple outline with a tight bounding box. Follow the provided examples and quality guidelines to ensure consistency.

• Inspect each image and locate purple-outlined characters (approximately 1–5 per image).
• Draw precise, non-rotated (axis-aligned) bounding boxes that tightly enclose each character.
• Ensure labels are consistent with examples and formatting required for YOLO training.
• Report ambiguous cases or quality issues through the project feedback channel.

Requirements

This is an intermediate-level labeling role: you should have direct experience creating object-detection bounding boxes for YOLO training and be comfortable using Roboflow or similar annotation tools.

• Proven experience labeling images for object detection to be used in a YOLO model (required).
• Attention to detail and ability to follow tight labeling guidelines.
• Reliable internet access and ability to work remotely.
• Available for up to 20 hours per week (part-time contractor).

Compensation & Schedule

This is contract, pay-per-label work. You are paid $0.02 USD per labeled object (per bounding box). You control how many batches you accept and can work up to, but less than, 20 hours per week.

• Payment type: Pay per label (PAY_PER_LABEL).
• Rate: $0.02 USD per labeled object.
• Employment: Contractor, part-time, worldwide and remote.

Who Should Apply

Apply if you have hands-on experience with image annotation for object detection (YOLO), are comfortable using Roboflow, want flexible remote work, and can maintain consistent, high-quality labels at scale.

• Intermediate-level labelers with YOLO experience will be prioritized.
• Good fit for freelancers, students, or part-time contributors seeking flexible schedules.

How It Works

After you apply, you'll receive a short qualification test and example images. Once approved, you can claim annotation batches in Roboflow, complete labels following the guide, and submit them for review.

Quality checks will be performed; consistent accuracy keeps you eligible for more batches.

• Qualification: Short test to confirm accuracy and tool familiarity.
• Workflow: Claim batches, annotate in Roboflow, submit for review, get paid per accepted label.