Annotator

Image annotator

Upwork • Freelance

May 2026 - Jul 2026 • 2 mos

I completed video annotation and action-captioning tasks for a confidential AI-training project involving short real-world activity videos. My role was to review each video, write a concise high-level summary, divide the footage into accurate action segments, and produce detailed, time-aligned captions in clear, simple present tense. A key challenge was ensuring complete coverage without gaps or overlaps while distinguishing similar objects, identifying the active hand, and avoiding assumptions about intent. I addressed this by reviewing videos at multiple speeds, fine-tuning segment boundaries frame by frame, verifying AI-generated caption suggestions against visible actions, and correcting inaccurate details. I also applied idle and invalid-action labels appropriately, reused consistent captions for repeated actions, and performed a final quality check on every task. This work strengthened my skills in video annotation, temporal segmentation, action captioning, AI-assisted review, guideline compliance, and dataset quality assurance.