Keywords: Instructional Videos, Task Graph, Keystep Recognition
TL;DR: We construct and use a task graph for keystep recognition. Contrary to the prior work, we use the longer activity context that improves state of the art performance on many instructional video datasets.
Abstract: Procedural activity understanding requires perceiving human actions in terms of a broader task, where multiple keysteps are performed in sequence across a long video to reach a final goal state---such as the steps of a recipe or the steps of a DIY fix-it task. Prior work largely treats keystep recognition in isolation of this broader structure, or else rigidly confines keysteps to align with a particular sequential script. We propose discovering a task graph automatically from how-to videos to represent probabilistically how people tend to execute keysteps, then leverage this graph to regularize keystep recognition in novel videos. On multiple datasets of real-world instructional video, we show the impact: more reliable zero-shot keystep localization and improved video representation learning, exceeding the state of the art.
Supplementary Material: pdf
Submission Number: 8401
Loading