Improving state estimation through projection post-processing for activity recognition with application to football

TL;DR

This paper introduces a post-processing technique to enhance activity recognition accuracy by correcting unrealistic short activities and proposes a new performance measure, validated on football sensor data and simulations.

Contribution

It presents a novel post-processing method for activity recognition and a new performance measure, the LTS measure, addressing timing uncertainties in state change detection.

Findings

The post-processing method improves classification accuracy.

The LTS measure effectively evaluates timing uncertainties.

Validated on both simulated and real football sensor data.

Abstract

The past decade has seen an increased interest in human activity recognition based on sensor data. Most often, the sensor data come unannotated, creating the need for fast labelling methods. For assessing the quality of the labelling, an appropriate performance measure has to be chosen. Our main contribution is a novel post-processing method for activity recognition. It improves the accuracy of the classification methods by correcting for unrealistic short activities in the estimate. We also propose a new performance measure, the Locally Time-Shifted Measure (LTS measure), which addresses uncertainty in the times of state changes. The effectiveness of the post-processing method is evaluated, using the novel LTS measure, on the basis of a simulated dataset and a real application on sensor data from football. The simulation study is also used to discuss the choice of the parameters of the post-processing method and the LTS measure.