Smartphone Screen Defect Segmentation

About OpenTrain AI

OpenTrain AI is the #1 platform for building careers in AI training and data labeling. We hire and contract contributors directly, give clear task instructions, and help you grow skills used to teach modern AI systems.

Creating an OpenTrain account is free. This role is contracted through OpenTrain AI and is part of real-world projects that improve how AI systems understand visual defects.

Why AI training matters

AI training (also called data labeling or annotation) is the human side of building intelligent systems. People prepare and review examples—like segmented defect masks—that models use to learn.

This type of work is flexible, remote, and often entry-friendly. Contributors help shape the behavior of computer vision systems used across manufacturing and refurbishment workflows.

The role

You will create pixel-accurate segmentation masks that highlight all surface defects on smartphone front-screen photos. The dataset contains 200 images, each a 20 megapixel photograph of a device front surface.

Pre-annotations are provided; your job is to refine and correct them so the annotated areas include all defect pixels and exclude background pixels as much as possible.

• Data type: high-resolution images (20 MP), 200 images total.
• Label type: segmentation (pixel-level masks).
• Tool: Label Studio (pre-annotations included).
• Defect examples to label: scratches, pit-defects, blemishes, missing surface areas, cracks, discolored areas.

What you'll do day-to-day

Work through assigned images in Label Studio, review provided pre-annotations, and correct or redraw segmentation masks to precisely cover defect pixels.

Follow annotation guidelines to differentiate defects from background reflections, fingerprints, or glare; ensure masks are tight around defects and avoid including non-defect pixels.

• Open and inspect each 20MP image and its pre-annotation.
• Refine polygon/mask annotations to include all defect pixels.
• Exclude background pixels and non-defect regions whenever possible.
• Submit completed annotations through Label Studio per project workflow.

Requirements

This is an entry-level role with no formal prerequisites beyond the project instructions. The listing specifies no other mandatory requirements.

The project is open worldwide and offered as a part-time contractor opportunity.

• Experience level: Entry level (no prior labeling experience required).
• Location: Remote — worldwide.
• Time commitment: Less than 20 hours per week (flexible).
• Employment type: Contractor, part-time.

Compensation & schedule

Pay is hourly at USD 2.00 per hour (PAY_PER_HOUR). Work is contractor, part-time, and flexible; you choose when to work as long as you meet project timelines and quality standards.

Because images are high-resolution, expect some tasks to take time for careful, pixel-accurate work.

• Hourly rate: $2.00 USD per hour.
• Workload: <20 hours/week, schedule up to you.
• Payment method and schedule will be handled by OpenTrain AI when you are contracted.

How to apply and start

Create an OpenTrain account (free), complete your profile, and apply to this project. If selected, you will receive access to the Label Studio workspace and the project annotation guidelines and pre-annotations.

Follow the provided instructions carefully and submit completed annotations through the Label Studio interface. OpenTrain supports contributors as they begin and while they work on projects.

• Sign up on OpenTrain and apply to this role.
• If accepted, you'll be given access to the dataset and Label Studio project.
• Work remotely, submit annotations, and receive payment through OpenTrain AI.