Smartphone Screen Defect Segmentation Task

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. Create a free account to discover projects, build your profile, and apply in minutes.

We help people start and grow careers teaching AI — by doing the hands-on work that shapes how models behave.

About AI training and image annotation

AI training (data labeling) is the human work behind modern machine learning systems. Image annotation — like the segmentation work in this project — is essential for teaching models to detect and understand visual defects.

This field is highly flexible and accessible: many projects require no prior experience, let you work remotely, and allow you to set hours that fit your schedule.

The role

You will create precise segmentation annotations that mark surface defects on smartphone front screens.

This is an entry-level, part-time contractor role. The dataset contains 200 high-resolution (20 megapixel) photos of smartphone front screens. Work is remote and open worldwide.

• Labeling type: Segmentation masks for defect regions.
• Tool: Label Studio (pre-annotations provided).
• Schedule: Less than 20 hours per week, flexible.
• Pay: $2.00 USD per hour (paid per hour).

What you'll do

Your main task is to review and edit segmentation masks so that every pixel belonging to a defect is included while excluding as much background as possible.

• Inspect each 20MP photo and confirm or correct pre-annotations provided in Label Studio.
• Mark all visible defects such as scratches, pit-defects, blemishes, missing surfaces, cracks, and discolored areas.
• Ensure annotation boundaries closely follow defect pixels and avoid including non-defect background.
• Follow project guidelines and maintain consistent labeling across the dataset.

Requirements

This project is entry level and requires attention to detail more than prior labeling experience. There are no specialized subject-matter prerequisites listed.

• Reliable internet connection and a computer capable of handling high-resolution images.
• Comfort using web-based annotation tools (Label Studio experience helpful but not required).
• Strong visual attention to detail and the ability to follow written labeling instructions.
• Available to work up to, but less than, 20 hours per week as a contractor.

Who should apply

Apply if you want flexible, remote part-time work contributing directly to how AI systems learn to detect hardware defects. This is a good fit for people new to data labeling who are careful, patient, and comfortable working with images.

How it works and how to apply

This is a contractor, part-time assignment paid hourly at $2 USD. You will work in Label Studio where pre-annotations are provided; your job is to correct and finalize segmentation masks according to the project guide.

To apply, create a free OpenTrain account, complete your profile, and submit your application to this project. OpenTrain connects you with the project and provides next-step instructions after selection.