Two-Stream Convolutional Networks for Action Recognition in Videos

TLDR

The challenge is to capture complementary appearance and motion information from still frames and inter‑frame motion. The study investigates deep ConvNet architectures for action recognition, seeks to generalise top hand‑crafted features into a data‑driven framework, and outlines three main contributions. They propose a two‑stream ConvNet with separate spatial and temporal networks, trained and evaluated on UCF‑101 and HMDB‑51 benchmarks. The architecture achieves strong performance, surpassing prior deep‑net approaches, and multi‑task learning across datasets further improves accuracy on UCF‑101 and HMDB‑51.

Abstract

We investigate architectures of discriminatively trained deep Convolutional Networks (ConvNets) for action recognition in video. The challenge is to capture the complementary information on appearance from still frames and motion between frames. We also aim to generalise the best performing hand-crafted features within a data-driven learning framework. Our contribution is three-fold. First, we propose a two-stream ConvNet architecture which incorporates spatial and temporal networks. Second, we demonstrate that a ConvNet trained on multi-frame dense optical flow is able to achieve very good performance in spite of limited training data. Finally, we show that multi-task learning, applied to two different action classification datasets, can be used to increase the amount of training data and improve the performance on both. Our architecture is trained and evaluated on the standard video actions benchmarks of UCF-101 and HMDB-51, where it is competitive with the state of the art. It also exceeds by a large margin previous attempts to use deep nets for video classification.