Towards Object-centric Understanding for Instructional Videos

Wenliang Guo, Yu Kong · 2025

Read Towards Object-centric Understanding for Instructional Videos by Wenliang Guo, Yu Kong (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-03479.txt.

Description

Videos that teach tasks need understanding beyond action labels. Wenliang Guo and Yu Kong propose an object-centric view in this 2025 work. Real procedures are flexible, and step order changes with the state of objects.

Action-centric methods struggle with that flexibility, the authors argue. They treat actions as mechanisms that drive state transitions of objects. Tracking objects and their states could help AI assistants follow recipes, repairs, and lab protocols. Such assistants must notice what changed, not just what motion occurred. Object states persist across steps, giving assistants a stable anchor in messy videos.

Catalog record

CREATORWenliang Guo, Yu Kong
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-03479

Browse the full library · About the Tower of Babel