Convolutional Two-Stream Network Fusion for Video Action Recognition

TL;DR

This paper introduces a novel convolutional two-stream network architecture that effectively fuses appearance and motion information at multiple layers for improved video action recognition, achieving state-of-the-art results.

Contribution

The paper presents a new ConvNet fusion architecture that combines spatial and temporal streams at convolutional layers, optimizing performance and efficiency for video action recognition.

Findings

Fusing at convolutional layers saves parameters without performance loss.

Late fusion at the last convolutional layer improves accuracy.

Pooling features over spatiotemporal neighborhoods enhances recognition.

Abstract

Recent applications of Convolutional Neural Networks (ConvNets) for human action recognition in videos have proposed different solutions for incorporating the appearance and motion information. We study a number of ways of fusing ConvNet towers both spatially and temporally in order to best take advantage of this spatio-temporal information. We make the following findings: (i) that rather than fusing at the softmax layer, a spatial and temporal network can be fused at a convolution layer without loss of performance, but with a substantial saving in parameters; (ii) that it is better to fuse such networks spatially at the last convolutional layer than earlier, and that additionally fusing at the class prediction layer can boost accuracy; finally (iii) that pooling of abstract convolutional features over spatiotemporal neighbourhoods further boosts performance. Based on these studies we propose a new ConvNet architecture for spatiotemporal fusion of video snippets, and evaluate its performance on standard benchmarks where this architecture achieves state-of-the-art results.