Subject

Gait analysis assesses the gait capabilities of a subject by quantifying kinematics (mostly joint angles) and kinetics (mostly joint moments) parameters. Spatiotemporal parameters (e.g. step length, step time and walking speed) are also popular in the rehabilitation field to assess for example the gait of elderly people and neurological patients (e.g. Stroke and Parkinson Disease).

Gait analysis based on marker based mocap - the state of the art - is performed using a biomechanical model, representing the human body as segments linked by balls joint in a chain. The orientation of the segments is determined by multiple cameras tracking the positions of infrared markers placed on the subject body. Marker based systems (e.g. Vicon) are therefore not portable and are difficult to use. Instead, the Kinect RGBD sensor offers a marker less, cheap, portable solution to track the subject body. Nevertheless, Kinect is limited by its inherent inaccuracy and the measured joint angles are generally not enough accurate for clinical assessment.

A potential solution to the limitations of both marker-based and markerless systems for gait analysis could be a hybrid approach that combines the advantages of both. By using the Kinect system in conjunction with markers, it's possible to maintain the portability and affordability of the Kinect while also achieving higher levels of precision. However, this approach requires an accurate estimation of marker positions from Kinect images, as well as the labeling of these markers to correctly assign them to specific positions on the body. This is where deep learning could come into play, as it has the potential to automate the marker labeling process and improve the accuracy of the hybrid approach. This project is focused on developing a deep learning model for automatic marker labeling, which could significantly enhance the effectiveness and accessibility of gait analysis for a wide range of applications.

The objective of this master's thesis is to explore and evaluate different deep learning approaches, such as transformers, for real-time automatic labeling of marker trajectories retrieved from Kinect images. Motion data from healthy volunteers will be used. The motion data were collected in the Brubotics Rehabilitation Research Center, equipped with Vicon and Kinect systems.

Kind of work

The project will consist of:
- Literature study.
- Obtain marker trajectories from Kinect images.
- training and testing different deep learning models on previous datasets of labeled marker trajectories.
- Extend the previous datasets by collecting new motion data (if needed).

The methods will be implemented in Python using open-source libraries and common deep learning frameworks. Data acquisition will require to operate the Vicon system (calibration, markers placement onto the subject body, use of the Vicon SDK software) and the Kinect system (Kinect SDK software). Note that Kinect SDK software requires a basic knowledge of c++.

Framework of the Thesis

The work will be done at VUB-ETRO, Pleinlaan 9 and VUB-BRRC (Laarbeeklaan 121, 1090 Brussels)

Number of Students

1