QC Reviewer - Human Pose Keypoint Annotations

About the Role

We are looking for a detail-oriented QC Reviewer to ensure the accuracy and consistency of human pose keypoint annotations on thermal images. You will review annotator submissions within Label Studio, verify correct keypoint placement (including inferred positions for occluded body parts), and provide clear feedback to drive rework where needed.

Key Responsibilities

• Review completed keypoint annotations against the source thermal images, verifying person count and per-person annotation completeness.
• Check that each keypoint is placed as close as possible to the true anatomical center of the corresponding body part.
• Evaluate whether inferred keypoints (for occluded/obstructed body parts) are anatomically plausible and consistent with visible body positioning.
• Verify that confidence flags were applied appropriately for uncertain or difficult cases, and flag missing or misused confidence indicators.
• Identify systematic errors or patterns of confusion (e.g., consistent mislabeling in certain poses or occlusion types) and report them for annotator calibration.
• Provide specific, actionable feedback on rejected or revised annotations within Label Studio's review workflow.
• Participate in periodic calibration sessions to align QC standards with the client's evolving guidelines or edge-case resolutions.
• Track and document review outcomes, common error types, and annotator performance trends as required.

Requirements

Required:

• Prior experience in computer vision annotation QA, review, or a related quality-control role.
• Strong understanding of human anatomy and pose structure to accurately assess keypoint placement, including inferred/occluded cases.
• Exceptional attention to detail — able to catch subtle placement errors, incorrect inferences, or misapplied confidence flags.
• Comfort working with thermal/infrared imagery and its visual limitations.
• Ability to apply detailed annotation/QC guidelines consistently and exercise sound judgment on ambiguous or edge-case scenarios.
• Strong written communication skills for providing clear, constructive feedback to annotators.
• Familiarity with Label Studio, including review/approval workflows, or willingness to quickly learn the platform.
• Reliable, self-directed work habits with the ability to meet review turnaround expectations.

Nice to Have:

• Direct experience with human pose estimation datasets (e.g., COCO keypoints) or keypoint QA specifically.
• Experience with thermal/infrared imagery QA.
• Background in computer vision, robotics, or security/surveillance imaging.
• Experience managing structured QA/rework feedback loops for annotation teams.