Comparative Analysis of CNN-based Spatiotemporal Reasoning in Videos

TL;DR

This paper compares various spatiotemporal modeling techniques for video action and gesture recognition, revealing that CNN-based methods outperform RNN-based approaches on benchmark datasets.

Contribution

It provides a comprehensive comparative analysis of different spatiotemporal modeling techniques applied to CNN features for video understanding tasks.

Findings

RNN-based techniques perform worse than convolutional architectures.

Fully convolutional models achieve better accuracy on benchmarks.

The study offers insights into effective spatiotemporal modeling for videos.

Abstract

Understanding actions and gestures in video streams requires temporal reasoning of the spatial content from different time instants, i.e., spatiotemporal (ST) modeling. In this survey paper, we have made a comparative analysis of different ST modeling techniques for action and gecture recognition tasks. Since Convolutional Neural Networks (CNNs) are proved to be an effective tool as a feature extractor for static images, we apply ST modeling techniques on the features of static images from different time instants extracted by CNNs. All techniques are trained end-to-end together with a CNN feature extraction part and evaluated on two publicly available benchmarks: The Jester and the Something-Something datasets. The Jester dataset contains various dynamic and static hand gestures, whereas the Something-Something dataset contains actions of human-object interactions. The common characteristic of these two benchmarks is that the designed architectures need to capture the full temporal content of videos in order to correctly classify actions/gestures. Contrary to expectations, experimental results show that Recurrent Neural Network (RNN) based ST modeling techniques yield inferior results compared to other techniques such as fully convolutional architectures. Codes and pretrained models of this work are publicly available.