A Closer Look at Spatiotemporal Convolutions for Action Recognition

TL;DR

This paper investigates various spatiotemporal convolutional methods for video action recognition, demonstrating that factorized 3D convolutions and the novel R(2+1)D block improve accuracy over traditional 2D CNNs.

Contribution

It introduces the R(2+1)D convolutional block, combining spatial and temporal filtering, achieving state-of-the-art results in action recognition datasets.

Findings

3D CNNs outperform 2D CNNs in accuracy.

Factorizing 3D convolutions improves performance.

R(2+1)D achieves comparable or superior results to state-of-the-art.

Abstract

In this paper we discuss several forms of spatiotemporal convolutions for video analysis and study their effects on action recognition. Our motivation stems from the observation that 2D CNNs applied to individual frames of the video have remained solid performers in action recognition. In this work we empirically demonstrate the accuracy advantages of 3D CNNs over 2D CNNs within the framework of residual learning. Furthermore, we show that factorizing the 3D convolutional filters into separate spatial and temporal components yields significantly advantages in accuracy. Our empirical study leads to the design of a new spatiotemporal convolutional block "R(2+1)D" which gives rise to CNNs that achieve results comparable or superior to the state-of-the-art on Sports-1M, Kinetics, UCF101 and HMDB51.