UntrimmedNets for Weakly Supervised Action Recognition and Detection

TL;DR

UntrimmedNet is a novel end-to-end architecture that learns action recognition and detection directly from untrimmed videos using weak supervision, reducing the need for costly temporal annotations.

Contribution

It introduces UntrimmedNet, a weakly supervised model with classification and selection modules, enabling effective learning from untrimmed videos for recognition and detection tasks.

Findings

Achieves comparable or superior performance to strongly supervised methods

Operates effectively without temporal annotations

Demonstrates success on THUMOS14 and ActivityNet datasets

Abstract

Current action recognition methods heavily rely on trimmed videos for model training. However, it is expensive and time-consuming to acquire a large-scale trimmed video dataset. This paper presents a new weakly supervised architecture, called UntrimmedNet, which is able to directly learn action recognition models from untrimmed videos without the requirement of temporal annotations of action instances. Our UntrimmedNet couples two important components, the classification module and the selection module, to learn the action models and reason about the temporal duration of action instances, respectively. These two components are implemented with feed-forward networks, and UntrimmedNet is therefore an end-to-end trainable architecture. We exploit the learned models for action recognition (WSR) and detection (WSD) on the untrimmed video datasets of THUMOS14 and ActivityNet. Although our UntrimmedNet only employs weak supervision, our method achieves performance superior or comparable to that of those strongly supervised approaches on these two datasets.