Pose is all you need: The pose only group activity recognition system (POGARS)

TL;DR

POGARS is a deep learning system that recognizes group activities solely from tracked poses, using 1D CNNs with attention mechanisms, achieving competitive results and better generalization than RGB-based methods.

Contribution

This paper introduces POGARS, a pose-only approach for group activity recognition utilizing 1D CNNs, spatial-temporal attention, and multi-task learning, which outperforms existing RGB-based methods.

Findings

Achieves state-of-the-art results on volleyball dataset

Demonstrates better generalization with pose-only input

Uses attention mechanisms for person-wise importance inference

Abstract

We introduce a novel deep learning based group activity recognition approach called the Pose Only Group Activity Recognition System (POGARS), designed to use only tracked poses of people to predict the performed group activity. In contrast to existing approaches for group activity recognition, POGARS uses 1D CNNs to learn spatiotemporal dynamics of individuals involved in a group activity and forgo learning features from pixel data. The proposed model uses a spatial and temporal attention mechanism to infer person-wise importance and multi-task learning for simultaneously performing group and individual action classification. Experimental results confirm that POGARS achieves highly competitive results compared to state-of-the-art methods on a widely used public volleyball dataset despite only using tracked pose as input. Further our experiments show by using pose only as input, POGARS has better generalization capabilities compared to methods that use RGB as input.