Improving Human Activity Recognition Through Ranking and Re-ranking

TL;DR

This paper introduces two ranking-based techniques, RaN and MIR, to improve human activity recognition accuracy by normalizing features and re-ranking classifier outputs, demonstrating significant performance gains on multiple datasets.

Contribution

The paper presents a novel parameter-free normalization method and a training-free re-ranking technique to enhance activity recognition systems.

Findings

RaN improves feature normalization for Fisher Vectors and VLAD.

MIR effectively re-ranks predictions by leveraging class relationships.

Methods significantly boost recognition accuracy on six datasets.

Abstract

We propose two well-motivated ranking-based methods to enhance the performance of current state-of-the-art human activity recognition systems. First, as an improvement over the classic power normalization method, we propose a parameter-free ranking technique called rank normalization (RaN). RaN normalizes each dimension of the video features to address the sparse and bursty distribution problems of Fisher Vectors and VLAD. Second, inspired by curriculum learning, we introduce a training-free re-ranking technique called multi-class iterative re-ranking (MIR). MIR captures relationships among action classes by separating easy and typical videos from difficult ones and re-ranking the prediction scores of classifiers accordingly. We demonstrate that our methods significantly improve the performance of state-of-the-art motion features on six real-world datasets.