Scene Cut Annotation — Temporal Segmentation for YouTube Videos

About OpenTrain

OpenTrain AI is the hiring and contracting organization for this project. OpenTrain is the #1 platform for finding and building careers in AI training and data labeling, connecting skilled contributors with meaningful annotation work across the industry.

We focus on projects that let people start and grow careers teaching AI. Creating an OpenTrain account is free and is the way to apply and manage your work if you’re selected.

About AI training and this type of work

AI training (also called data labeling or annotation) is the human work that teaches models how to understand and edit media. Video temporal segmentation — marking precise scene boundaries — is essential for summarization, automated editing, and many downstream video applications.

This type of work is often remote and flexible, accessible to people with video-editing or annotation experience, and gives contributors a direct role in shaping how modern AI systems handle video content.

Project overview

OpenTrain AI is contracting a vendor or team to produce high-quality temporal segmentation labels for a large batch of YouTube videos. Annotators will mark scene boundaries (cuts or transitions) and record start/end timestamps for each scene according to provided guidelines.

Scope: annotate scene boundaries in provided videos, mark start and end timestamps per scene, follow detailed guidelines and examples, perform internal quality checks and revisions, and deliver final output in JSON or CSV.

• Data type: Video — temporal segmentation (scene boundary annotation).
• Volume: ~1,000 videos (average 5–10 minutes each).
• Annotation tool: uLabel (access will be provided).
• Expected output: JSON/CSV with fields: video_id, scene_id, start_time, end_time.

What you'll do

Work as part of a vendor team or annotator group to apply temporal segmentation labels to assigned videos. Follow the project’s detailed annotation guidelines and use the uLabel tool for marking timestamps.

Conduct internal QA on your work before delivery, respond to reviewer feedback, and meet weekly quality audits from the client’s QA team.

• Identify cuts and transitions and mark precise scene start/end timestamps.
• Use uLabel and export annotations to the required JSON/CSV schema.
• Perform internal QC and revise annotations prior to submission.
• Meet agreed turnaround timelines and communicate capacity changes promptly.

Quality, timelines, and team size

This project requires high accuracy and fast, reliable delivery. Quality targets and scheduling are firm and will be enforced through weekly audits.

Preferred vendor/team size and project duration are provided to help bidders plan capacity and delivery.

• Quality target: ≥95% accuracy on spot checks.
• Weekly quality audits by the internal QA team.
• Preferred team size: 10–15 annotators.
• Project duration: 4–6 weeks.
• Must meet agreed turnaround timelines.

Requirements & qualifications

Candidates should be intermediate-level annotators or vendors with proven experience in video annotation or temporal segmentation. You must be able to run an internal QA process and demonstrate capacity to deliver at scale.

Open to worldwide applicants; work will be coordinated and contracted through OpenTrain AI.

• Prior experience with video annotation or temporal segmentation (required).
• Familiarity with timestamp-based scene marking and exporting JSON/CSV outputs.
• Ability to run internal QA and achieve ≥95% accuracy on spot checks.
• Capacity to staff a 10–15 person team or equivalent throughput for the project duration.
• Contractor engagement: you will be engaged as a contractor for the project.

Proposal & pricing instructions

Submit a proposal that demonstrates your experience, capacity, timeline, and quality processes. The posting provides the project pay rate and asks bidders to confirm how they will meet delivery and QA targets.

Compensation: the project pays $4.00 USD per hour (PAY_PER_HOUR). In your proposal, also provide your preferred pricing breakdown (per video or per hour of video) to help evaluate operational costs and scheduling.

• Include prior video annotation or temporal-segmentation examples and references.
• State the estimated team size and daily/weekly delivery capacity.
• Provide cost estimates (cost per annotated video or per hour of video annotated) in addition to acknowledging the $4/hr pay rate.
• Detail your internal QA process, accuracy monitoring, and revision workflow.
• Confirm ability to meet a 4–6 week schedule and weekly audit cadence.

How to apply

Prepare a concise proposal that addresses the items above. OpenTrain AI will review submissions and invite selected vendors/teams to onboarding and tool access.

If you do not yet have an OpenTrain account, sign up (it’s free) to submit your proposal and manage any follow-up questions.

• Submit: short company/team summary, relevant experience, sample work, proposed team size, delivery timeline, pricing breakdown, and QA approach.
• Be ready to start onboarding within the project timeline and accept uLabel access for annotation.
• Applications will be assessed for quality processes, demonstrated throughput, and ability to meet the ≥95% accuracy target.