Polygon Segmentation for Recycling Images

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. It connects people with projects that teach and shape AI — discover open projects, build a profile, and apply in minutes; creating an account is free.

Why AI Training Matters

AI training (also called data labeling or annotation) is the human work behind how models learn: people prepare, correct, and review examples so systems make better decisions. These roles are often fully remote, flexible, and accessible — many projects need attention to detail more than prior experience, and contributors directly influence how AI behaves.

The Role

We need contributors to create polygon segmentations for recycling images using CVAT together with the Segment Anything interactor. Objects above a specified size threshold must be outlined with polygons that distinguish individual items in frames containing multiple objects.

This is an entry-level contractor role, open worldwide and paid at a fixed rate of $5.00 USD per hour. Work begins with a small paid sample so you can demonstrate that you meet the project's speed and accuracy expectations; qualifying contributors may receive additional assignments.

• Data type: images
• Label types: object detection, polygon segmentation
• Labeling software: CVAT (using the Segment Anything interactor alongside CVAT)
• Employment type: contractor; worldwide applicants accepted

What You'll Do

Follow project guidelines to create precise polygon annotations for individual recycling items. When multiple objects appear in a single frame, clearly separate each item with its own polygon if it exceeds the size threshold.

• Open images in CVAT and use the Segment Anything interactor to assist segmentation where allowed.
• Draw polygon outlines that accurately follow object boundaries and label distinct items separately.
• Adhere to size-threshold rules: only objects above the specified threshold require polygon annotation.
• Work to meet the project's accuracy and speed requirements; reviewers will evaluate submissions.

Requirements

This is an entry-level task but attention to detail and the ability to distinguish individual objects in cluttered frames are essential. Familiarity with CVAT and the Segment Anything interactor is advantageous but not strictly required.

• Able to use CVAT for polygon annotation; experience with the Segment Anything interactor is a plus.
• Strong visual attention to detail and the ability to separate overlapping or adjacent items.
• Reliable internet access and a computer capable of running CVAT.
• Comfort working as a remote contractor and accepting hourly pay ($5.00 USD/hr).

How It Works

You will start with a small paid batch so the team can assess speed and accuracy. If your work meets the project's standards, additional tasks may be offered.

Submit annotations according to the project's QC process and feedback. Payment is hourly in USD at the stated rate for contractor work.

• Small paid trial → performance review → potential additional assignments.
• Maintain consistent quality and follow reviewer feedback to increase opportunities.

Who Should Apply

Apply if you want remote, flexible annotation work and can make precise polygon masks in images. This role suits people new to AI training who have good visual discrimination and are willing to learn project rules and tooling.

• Entry-level applicants welcome.
• Good fit for detail-oriented contributors who want to grow in AI labeling work.