Polygon Segmentation For Game Screenshots

About OpenTrain AI

OpenTrain AI is the hiring and contracting organization for this role. OpenTrain is the #1 platform for starting and building careers in AI training and data labeling. Creating an OpenTrain account is free.

• Work with a leading AI-training organization and gain experience in data annotation.
• Open, remote roles that help contributors build a profile and find more projects in the industry.

About AI Training Work

AI training (also called data labeling or annotation) is the human work that teaches machine learning models how to perceive and act. Contributors annotate images, audio, and text to create examples that models learn from.

This kind of work is often fully remote, flexible, and accessible to entry-level contributors. Projects like this one put you on the cutting edge of how game visuals and interactive tasks get represented to AI systems.

• Flexible, remote work you can do from anywhere.
• Entry-level projects are common — domain experience (like gaming) can help you move faster.

The Role

Label polygon shapes that represent all acceptable click points for an objective in game screenshots. The dataset is approximately 1,000 images of a house scene (door, garden, tree) with text prompts like "Click on the door."

Your output for each image will be a polygon that covers the door area — i.e., all acceptable points where a click would fulfill the prompt. Work will be done inside Labelbox.

• Project size: ~1,000 images of the same scene type.
• Data type: IMAGE; label type: POLYGON.
• Labeling tool: LABELBOX.

What You'll Do

Perform polygon segmentation on screenshots to mark the clickable door area that satisfies the instruction "Click on the door." Each polygon should include all acceptable click points for the objective.

Follow the per-image instruction and submit completed polygons through the Labelbox interface.

• Open assigned screenshots in Labelbox and draw polygons around the door.
• Ensure polygons cover all acceptable click locations for the task prompt.
• Work at a steady, consistent pace to complete the dataset.

Requirements

This is an entry-level contract role. There is no formal prior experience required; contributors with some experience playing video games may find the task easier to understand.

The project accepts contributors worldwide and is offered as a part-time contractor role. Work will be completed in Labelbox.

• Experience level: Entry level.
• Helpful but not required: prior experience playing video games.
• Worldwide applicants accepted; role is remote and contract/part-time.

Compensation & Time Commitment

Pay is hourly at USD 8.00 per hour. This role is billed pay-per-hour.

Expected time commitment is less than 20 hours per week; schedule is flexible within that cap.

• Payment type: Pay per hour — $8.00 USD/hr.
• Time requirement: Less than 20 hours/week.

How It Works

Apply through your OpenTrain account (creating one is free). If selected, you'll receive access to the Labelbox workspace and the assigned images to label.

Work remotely in Labelbox, draw polygons per instruction, and submit labels for review. Follow any project-specific guidance provided in the Labelbox task UI.

• Create an OpenTrain account and submit your application to this project.
• Label tasks are completed in Labelbox and submitted through that tool.

How To Apply

Create a free OpenTrain account, complete your profile, and apply to this project. Make sure your profile notes that you can work part-time as a contractor and that you can use Labelbox if asked.

Applications are reviewed by the OpenTrain AI team; accepted contributors will be given access and instructions inside Labelbox.

• OpenTrain account required (free).
• Apply from your OpenTrain profile and indicate availability under 20 hours/week.