Learning zeroth class dictionary for human action recognition

TL;DR

This paper introduces a two-phase dictionary learning approach for human action recognition that includes a novel 'zeroth class' method to filter out non-discriminative frames, improving recognition accuracy.

Contribution

It presents a new discriminative two-phase dictionary learning framework with a 'zeroth class' trick for better handling of non-informative frames in action recognition.

Findings

Effective detection of non-discriminative frames

Improved recognition accuracy on benchmark datasets

Demonstrated robustness of the method

Abstract

In this paper, a discriminative two-phase dictionary learning framework is proposed for classifying human action by sparse shape representations, in which the first-phase dictionary is learned on the selected discriminative frames and the second-phase dictionary is built for recognition using reconstruction errors of the first-phase dictionary as input features. We propose a "zeroth class" trick for detecting undiscriminating frames of the test video and eliminating them before voting on the action categories. Experimental results on benchmarks demonstrate the effectiveness of our method.