University of Twente Student Theses


Overground vs. treadmill: where is my horse running? Context detection based on IMU data

Xu, Zihao (2022) Overground vs. treadmill: where is my horse running? Context detection based on IMU data.

This is the latest version of this item.

[img] PDF
Abstract:Horse gaits and lameness can be classified and studied either in-hand overground or on a treadmill, using, for example, optical motion capture (OMC) or inertial measurement unit (IMU) systems. Therefore, the context of horse locomotion bears particular importance and value in the field of research. Especially, treadmill evaluations enable gathering highly standardized data at a steady state. Research on developing classification models for overground vs. treadmill locomotion is solicited to improve the automation of an existing gait analysis system. This research will train respectively three feature-based machine learning (ML) and two signal-based deep learning (DL) models by the collected data, with IMUs mounted on both upper-body (poll, withers, pelvis, see in Figure 1) and limbs, from overground and treadmill terrain types. The data are then divided into three datasets, followed by an analysis evaluating different approaches’ performances on each set. This research reveals that ML models generally achieve considerably higher accuracy and stability than DL models. Besides, ML models applied with the feature reduction technique experienced a performance drop.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science BSc (56964)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page