A Pursuit of Temporal Accuracy in General Activity Detection

TL;DR

This paper introduces a new framework for more accurate temporal activity detection in untrimmed videos, addressing challenges in locating activity boundaries and distinguishing relevant segments.

Contribution

It presents a novel candidate proposal scheme and a cascaded classification pipeline, improving boundary accuracy and relevance assessment for diverse activities.

Findings

Significantly outperforms existing methods on THUMOS14 and ActivityNet datasets.

Demonstrates superior accuracy and adaptability across various activity temporal structures.

Provides a generic framework applicable to multiple activity detection scenarios.

Abstract

Detecting activities in untrimmed videos is an important but challenging task. The performance of existing methods remains unsatisfactory, e.g., they often meet difficulties in locating the beginning and end of a long complex action. In this paper, we propose a generic framework that can accurately detect a wide variety of activities from untrimmed videos. Our first contribution is a novel proposal scheme that can efficiently generate candidates with accurate temporal boundaries. The other contribution is a cascaded classification pipeline that explicitly distinguishes between relevance and completeness of a candidate instance. On two challenging temporal activity detection datasets, THUMOS14 and ActivityNet, the proposed framework significantly outperforms the existing state-of-the-art methods, demonstrating superior accuracy and strong adaptivity in handling activities with various temporal structures.