Data labeling
VideoText Summarization
Egocentric Annotation: Performed high-precision temporal labeling of manual tasks for AI human-object interaction models.
Taxonomy & Flow: Applied strict "Verb + Object" framing with zero-gap transitions between action sequences.
Logic Modeling: Classified "Failures" and "Recoveries" to improve model accuracy in task success recognition.
Quality Assurance: Conducted data validation by flagging clips based on focus, framing, and movement integrity.
Egocentric Annotation: Performed high-precision temporal labeling of manual tasks for AI human-object interaction models. Taxonomy & Flow: Applied strict "Verb + Object" framing with zero-gap transitions between action sequences. Logic Modeling: Classified "Failures" and "Recoveries" to improve model accuracy in task success recognition. Quality Assurance: Conducted data validation by flagging clips based on focus, framing, and movement integrity.
2026 - Present