Revisiting hand-crafted feature for action recognition: a set of improved dense trajectories

TL;DR

This paper introduces Trajectory-Set, a new feature for action recognition based on improved Dense Trajectories that focuses on trajectory information alone, achieving competitive results without appearance features.

Contribution

It presents a novel trajectory-based feature called Trajectory-Set that enhances action recognition performance by focusing on motion trajectories without appearance data.

Findings

TS outperforms many existing methods on UCF50, UCF101, and HMDB51 datasets.

Achieves 85.4% accuracy on HMDB51, surpassing some deep learning approaches.

Code is publicly available for reproducibility.

Abstract

We propose a feature for action recognition called Trajectory-Set (TS), on top of the improved Dense Trajectory (iDT). The TS feature encodes only trajectories around densely sampled interest points, without any appearance features. Experimental results on the UCF50, UCF101, and HMDB51 action datasets demonstrate that TS is comparable to state-of-the-arts, and outperforms many other methods; for HMDB the accuracy of 85.4%, compared to the best accuracy of 80.2% obtained by a deep method. Our code is available on-line at https://github.com/Gauffret/TrajectorySet .