Tier 3 Reviewer (T3)
VideoText Generation
There are various episodes (full video tasks) and each episode contains several segments (a continuous time span paired with one label). My task is to generate text that accurately describes the major atomic actions performed by the ego (human) in the video. Part of the quality measures deployed is non-usage of certain verbs, strict adherence to time spans of 10 seconds maximum per segment and many more.
There are various episodes (full video tasks) and each episode contains several segments (a continuous time span paired with one label). My task is to generate text that accurately describes the major atomic actions performed by the ego (human) in the video. Part of the quality measures deployed is non-usage of certain verbs, strict adherence to time spans of 10 seconds maximum per segment and many more.
2026 - Present