Video Data Labeling
VideoSegmentation
I contributed to a hand-label segmentation masks frame-by-frame for objects and scenes to train computer vision models. Responsibilities included accurately drawing polygon masks around objects, tracking consistency of objects between frames, addressing occlusion and motion changes per strict guidelines. Worked on over 2,000 video clips (150,000+ frames). Accuracy was validated through multi-pass QC, inter-annotator agreement samples, and automated sanity checks. Achieved >95% segmentation accuracy and robust temporal consistency metrics.
I contributed to a hand-label segmentation masks frame-by-frame for objects and scenes to train computer vision models. Responsibilities included accurately drawing polygon masks around objects, tracking consistency of objects between frames, addressing occlusion and motion changes per strict guidelines. Worked on over 2,000 video clips (150,000+ frames). Accuracy was validated through multi-pass QC, inter-annotator agreement samples, and automated sanity checks. Achieved >95% segmentation accuracy and robust temporal consistency metrics.
2024 - Present