Data Annotation & Labelling

We are looking for a detail-oriented data annotation specialist or annotation team to help review and improve AI-generated captions and temporal action segments for first-person industrial/task videos. The videos show hands performing step-by-step tasks such as assembly, food preparation, ironing, lathe/marking work, and similar manual workflows. Our AI system generates initial segment boundaries and captions, and we need human reviewers to verify and correct them. Main Work You will review short video clips and ensure that each action segment is accurate. - For each segment, you will check or correct: - Start and end timestamps - Detailed description of what both hands are doing - Task phase label, such as pick, place, align, mark, stir, wipe, adjust - Object visible in the left hand - Object visible in the right hand - Relevant visible objects in the scene Segments must represent meaningful sub-actions. For example, if a person picks up a part and places it into a bin, that full pick-to-place action should usually be one segment. Additional Data Annotation Work We also need support for visual annotation tasks, including: - Bounding boxes around hands, tools, objects, containers, and work surfaces - Hand mesh or hand pose annotation where required - Object tracking across frames - Quality control of annotations - Consistent labeling across multiple videos and task types Experience with CVAT, Label Studio, Supervisely, Roboflow, or similar annotation tools is preferred. Ideal Candidate You should be comfortable with: - Watching detailed hand-object interaction videos - Making precise frame-level timestamp decisions - Writing clear English captions - Following strict annotation guidelines - Creating accurate bounding boxes - Working with hand pose, hand mesh, or keypoint annotation workflows - Handling repetitive review tasks carefully and consistently Requirements Strong attention to detail Good written English Prior experience with video annotation or computer vision datasets Ability to follow annotation guidelines exactly Availability for an initial paid trial task Team capacity is a plus, since this may scale to a larger dataset Trial Task We will provide a small sample video and annotation guidelines. The trial will involve reviewing AI-generated action segments, correcting captions, and optionally annotating hands/objects depending on your skill set. Please apply with: Examples of previous annotation work Tools you have used Whether you can do bounding boxes, hand pose/mesh, or both Your hourly rate or per-video/per-frame pricing Your team size and weekly capacity, if applicable