Journal
DISPLAYS
Volume 78, Issue -, Pages -Publisher
ELSEVIER
DOI: 10.1016/j.displa.2023.102434
Keywords
Temporal action localization; Active learning; Scoring function
Ask authors/readers for more resources
This paper introduces an active learning method called AL-STAL to address the issue of high labeling cost in action localization tasks. By selecting video samples with high informativeness and training the localization model, AL-STAL achieves superior performance compared to existing methods and demonstrates satisfying performance compared to fully-supervised learning on three benchmark datasets.
Temporal Action Localization (TAL) aims to predict both action category and temporal boundary of action instances in untrimmed videos, i.e., start and end time. Existing works usually adopt fully-supervised solutions, however, one of the practical bottlenecks in these solutions is the large amount of labeled training data required. To reduce expensive human label cost, this paper focuses on a rarely investigated yet practical task named semi-supervised TAL and proposes an effective active learning method, named AL-STAL. We leverage four steps for actively selecting video samples with high informativeness and training the localization model, named Train, Query, Annotate, Append. Two scoring functions that consider the uncertainty of localization model are equipped in AL-STAL, thus facilitating the video sample ranking and selection. One takes entropy of predicted label distribution as measure of uncertainty, named Temporal Proposal Entropy (TPE). And the other introduces a new metric based on mutual information between adjacent action proposals, named Temporal Context Inconsistency (TCI). To validate the effectiveness of proposed method, we conduct extensive experiments on three benchmark datasets THUMOS'14, ActivityNet 1.3 and ActivityNet 1.2. Experiment results show that AL-STAL outperforms the existing competitors and achieves satisfying performance compared with fully-supervised learning.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available