AI Annotator
The aim of the project was to train AI to see motion as story unfolding not as blur of pixels, The video annotated often involved; Real world environments, Human movements and gestures, and complex multi object interaction. I was tasked with; drawing bounding boxes around objects, Labeling object class, defining event duration, maintaing chronological intergrity. Large scale data set were provided from all over social media to millions of annotators, reviewers and QA specialists. The following are some of the quality guidelines we adhered to; Following detailed annotation class taxonomies, Identifying variance trends, accuracy scoring systems.