Improving state estimation through projection post-processing for activity recognition with application to football
This work addresses the problem of improving the accuracy of activity recognition and its evaluation for researchers and practitioners working with sensor data, particularly in sports analytics; it is an incremental improvement.
This paper introduces a post-processing method to improve activity recognition by correcting for unrealistic short activities in state estimates. It also proposes a new performance measure, the Locally Time-Shifted Measure (LTS measure), to address uncertainty in state change times, evaluating both on simulated data and real football sensor data.
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.