Optical Flow Guided Feature: A Fast and Robust Motion Representation for Video Action Recognition

TL;DR

This paper introduces Optical Flow Guided Feature (OFF), a fast, robust, and compact motion representation for video action recognition that improves accuracy while significantly reducing computational cost.

Contribution

The paper proposes a novel OFF method derived from optical flow principles, enabling efficient spatiotemporal feature extraction in CNNs for action recognition.

Findings

Achieves 93.3% accuracy on UCF-101 with only RGB inputs.

OFF is 15 times faster than two-stream methods.

Complementary to optical flow, improving overall accuracy.

Abstract

Motion representation plays a vital role in human action recognition in videos. In this study, we introduce a novel compact motion representation for video action recognition, named Optical Flow guided Feature (OFF), which enables the network to distill temporal information through a fast and robust approach. The OFF is derived from the definition of optical flow and is orthogonal to the optical flow. The derivation also provides theoretical support for using the difference between two frames. By directly calculating pixel-wise spatiotemporal gradients of the deep feature maps, the OFF could be embedded in any existing CNN based video action recognition framework with only a slight additional cost. It enables the CNN to extract spatiotemporal information, especially the temporal information between frames simultaneously. This simple but powerful idea is validated by experimental results. The network with OFF fed only by RGB inputs achieves a competitive accuracy of 93.3% on UCF-101, which is comparable with the result obtained by two streams (RGB and optical flow), but is 15 times faster in speed. Experimental results also show that OFF is complementary to other motion modalities such as optical flow. When the proposed method is plugged into the state-of-the-art video action recognition framework, it has 96:0% and 74:2% accuracy on UCF-101 and HMDB-51 respectively. The code for this project is available at https://github.com/kevin-ssy/Optical-Flow-Guided-Feature.