Spot On: Action Localization from Pointly-Supervised Proposals

TL;DR

This paper introduces a point-based annotation method for action localization in videos, reducing annotation effort while maintaining competitive accuracy compared to traditional box annotations.

Contribution

It proposes a novel point-based annotation approach combined with a MIL framework for efficient spatio-temporal action localization in videos.

Findings

Point annotations are faster to produce than box annotations.

The approach achieves comparable localization performance to box-based methods.

It is effective with minimal supervision, competitive with state-of-the-art methods.

Abstract

We strive for spatio-temporal localization of actions in videos. The state-of-the-art relies on action proposals at test time and selects the best one with a classifier trained on carefully annotated box annotations. Annotating action boxes in video is cumbersome, tedious, and error prone. Rather than annotating boxes, we propose to annotate actions in video with points on a sparse subset of frames only. We introduce an overlap measure between action proposals and points and incorporate them all into the objective of a non-convex Multiple Instance Learning optimization. Experimental evaluation on the UCF Sports and UCF 101 datasets shows that (i) spatio-temporal proposals can be used to train classifiers while retaining the localization performance, (ii) point annotations yield results comparable to box annotations while being significantly faster to annotate, (iii) with a minimum amount of supervision our approach is competitive to the state-of-the-art. Finally, we introduce spatio-temporal action annotations on the train and test videos of Hollywood2, resulting in Hollywood2Tubes, available at http://tinyurl.com/hollywood2tubes.