JARViS: Detecting Actions in Video Using Unified Actor-Scene Context Relation Modeling

TL;DR

JARViS introduces a novel two-stage video action detection framework that leverages unified actor-scene context modeling with Transformer attention, significantly improving performance over existing methods.

Contribution

The paper presents JARViS, a new two-stage VAD framework that effectively models cross-modal actor-scene relations using Transformer attention, achieving state-of-the-art results.

Findings

Outperforms existing VAD methods on AVA, UCF101-24, JHMDB51-21 datasets.

Achieves significant performance improvements with Transformer-based context modeling.

Demonstrates the effectiveness of unified actor-scene relation modeling in video action detection.

Abstract

Video action detection (VAD) is a formidable vision task that involves the localization and classification of actions within the spatial and temporal dimensions of a video clip. Among the myriad VAD architectures, two-stage VAD methods utilize a pre-trained person detector to extract the region of interest features, subsequently employing these features for action detection. However, the performance of two-stage VAD methods has been limited as they depend solely on localized actor features to infer action semantics. In this study, we propose a new two-stage VAD framework called Joint Actor-scene context Relation modeling based on Visual Semantics (JARViS), which effectively consolidates cross-modal action semantics distributed globally across spatial and temporal dimensions using Transformer attention. JARViS employs a person detector to produce densely sampled actor features from a keyframe. Concurrently, it uses a video backbone to create spatio-temporal scene features from a video clip. Finally, the fine-grained interactions between actors and scenes are modeled through a Unified Action-Scene Context Transformer to directly output the final set of actions in parallel. Our experimental results demonstrate that JARViS outperforms existing methods by significant margins and achieves state-of-the-art performance on three popular VAD datasets, including AVA, UCF101-24, and JHMDB51-21.