University of Twente Student Theses
Classification of eating gestures using a wrist worn IMU and the deep learning model InceptionTime.
Loh, Sönke van (2022) Classification of eating gestures using a wrist worn IMU and the deep learning model InceptionTime.
|
PDF
1MB |
| Abstract: | This paper deals with the classification of eating related gestures using 3 axis accelerometer and gyroscope data from a wrist worn Inertial Measurement Unit (IMU) for the purpose of dietary monitoring. The data is used to train the convolutional neural network (CNN) InceptionTime. It is gathered in an experiment consisting of 9 participants and contains 8 classes which the network needs to classify. The data is fed to the network as a multivariate time series (MTS) which means that the 6 different channels of measurements are treated as one time series in classification. The results of the experiments are compared to results from the master thesis of Sigert Mevissen as an extension of his work. It is confirmed that the approach of using a CNN for classification of MTS is applicable in the case of eating gesture recognition. On a set containing all 9 participants, 69% F1 score is achieved. When combining the eating and non-eating gestures into a binary classification, this increases to 80%. In three leave one subject out (loso) tests, F1 score of 63% on average are achieved. The SVM from the master thesis, trained on another dataset, achieved 82% F1 score on a full and 18% F1 score on loso test sets. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 50 technical science in general |
| Programme: | Electrical Engineering BSc (56953) |
| Link to this item: | https://purl.utwente.nl/essays/92323 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page