Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Graph-based Approach to Human Activity Recognition

Thomas Peroutka, Ilir Murturi, Praveen Kumar Donta, Schahram Dustdar

TL;DR

This work tackles real-time human activity recognition from large multi-sensor streams by encoding expert knowledge of movements into directed graphs and recognizing sequences via a deterministic automaton. By decomposing complex athletic movements into points of interest and using generic and sensor-specific triggers on IMU/GNSS data, the method yields explainable, efficient detection and metrics extraction, demonstrated on biathlon data. The approach supports combining partial solutions to form total solutions and optimizes them by duration and quantity, enabling robust analysis of performance across laps, penalties, and shooting phases. Its resource-efficient, explainable framework offers a practical alternative to deep learning for large-scale HAR with potential for real-time, edge-enabled analytics in sports and rehabilitation.

Abstract

Advanced wearable sensor devices have enabled the recording of vast amounts of movement data from individuals regarding their physical activities. This data offers valuable insights that enhance our understanding of how physical activities contribute to improved physical health and overall quality of life. Consequently, there is a growing need for efficient methods to extract significant insights from these rapidly expanding real-time datasets. This paper presents a methodology to efficiently extract substantial insights from these expanding datasets, focusing on professional sports but applicable to various human activities. By utilizing data from Inertial Measurement Units (IMU) and Global Navigation Satellite Systems (GNSS) receivers, athletic performance can be analyzed using directed graphs to encode knowledge of complex movements. Our approach is demonstrated on biathlon data and detects specific points of interest and complex movement sequences, facilitating the comparison and analysis of human physical performance.

A Graph-based Approach to Human Activity Recognition

TL;DR

This work tackles real-time human activity recognition from large multi-sensor streams by encoding expert knowledge of movements into directed graphs and recognizing sequences via a deterministic automaton. By decomposing complex athletic movements into points of interest and using generic and sensor-specific triggers on IMU/GNSS data, the method yields explainable, efficient detection and metrics extraction, demonstrated on biathlon data. The approach supports combining partial solutions to form total solutions and optimizes them by duration and quantity, enabling robust analysis of performance across laps, penalties, and shooting phases. Its resource-efficient, explainable framework offers a practical alternative to deep learning for large-scale HAR with potential for real-time, edge-enabled analytics in sports and rehabilitation.

Abstract

Advanced wearable sensor devices have enabled the recording of vast amounts of movement data from individuals regarding their physical activities. This data offers valuable insights that enhance our understanding of how physical activities contribute to improved physical health and overall quality of life. Consequently, there is a growing need for efficient methods to extract significant insights from these rapidly expanding real-time datasets. This paper presents a methodology to efficiently extract substantial insights from these expanding datasets, focusing on professional sports but applicable to various human activities. By utilizing data from Inertial Measurement Units (IMU) and Global Navigation Satellite Systems (GNSS) receivers, athletic performance can be analyzed using directed graphs to encode knowledge of complex movements. Our approach is demonstrated on biathlon data and detects specific points of interest and complex movement sequences, facilitating the comparison and analysis of human physical performance.
Paper Structure (21 sections, 4 equations, 20 figures, 1 table, 2 algorithms)

This paper contains 21 sections, 4 equations, 20 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Motivation and Related Work
  3. Motivation Scenario
  4. Related Work
  5. The proposed approach
  6. An overview of the approach
  7. Encoding complex movements into a directed graph
  8. Detection of points of interest in multi-sensor datasets
  9. Generic triggers
  10. Sensor-specific triggers
  11. A direct graph to recognize complex movements
  12. Combine partial solutions to find multiple total solutions
  13. Find optimal total solution
  14. Maximize number of partial solutions
  15. Maximize the covered duration
  16. ...and 6 more sections

Figures (20)

  • Figure 1: An overview of the proposed approach.
  • Figure 2: A typical biathlon racetrack layout.
  • Figure 3: Acceleration data from an uphill section.
  • Figure 4: Macro-view of biathlon. Turning sequence of events into a directed graph.
  • Figure 8: Virtual gates track layout.
  • ...and 15 more figures