SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.13183 · CVPR · 2021

The Blessings of Unlabeled Background in Untrimmed Videos

Hanwang Zhang, Jianqiang Huang, Yuan Liu, Zhenfang Chen, Jingyuan Chen, Bing Deng

arXiv · PDF · Open in the Atlas

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. If it links a repository, it is listed below.

Abstract

The key challenge is how to distinguish the action of interest segments from the background, which is unlabelled even on the video-level. While previous works treat the background as "curses", we consider it as "blessings". Specifically, we first use causal analysis to point out that the common localization errors are due to the unobserved confounder that resides ubiquitously in visual recognition. Then, we propose a Temporal Smoothing PCA-based (TS-PCA) deconfounder, which exploits the unlabelled background to model an observed substitute for the unobserved confounder, to remove the confounding effect. Note that the proposed deconfounder is model-agnostic and non-intrusive, and hence can be applied in any WTAL method without model re-designs. Through extensive experiments on four state-ofthe-art WTAL methods, we show that the deconfounder can improve all of them on the public datasets: THUMOS-14 and ActivityNet-1.3 1 .

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2103.13183")
get_code_for_paper("2103.13183")
have("2103.13183")

Connect an agent — have() is free.