Seminar: Automatic Labelling for Chewing Timeseries

Symbolbild zum Artikel. Der Link öffnet das Bild in einer großen Anzeige.
Example Temporalis chewing EMG signals. All Chewing cycles were labelled manually. An automatic labelling tool can release the human workload.

Course/Project description

Chewing is an essential component of eating. Chewing cycles can be recorded using various sensors and detected applying appropriate algorithms. However, evaluating a chewing detection algorithm is usually difficult due to the lack of accurate ground truth. It is even more difficult to label free-living data of a tremendous number of chewing cycle instances. Based on our investigation, Temporalis electromyography (EMG) can serve as a reliable source of ground truth among a few other choices, e.g. videos and dietary diaries. The proposed student project aims at providing an accurate automatic chewing cycle labelling tool for EMG signals. The onsets and offsets of chewing cycles shall be detected and labelled. The labelling tool will be implemented in Python. The automatic labelling performance will be also evaluated compared with manual labels.


Learning objectives

  • Python and graphical programming
  • Timeseries processing and chewing detection

Course data

ECTS 2.5, 5, 7.5, default: 5
Project type BSc./MSc.-Seminar, Thesis
Language English
Presence time 4 SWS
Useful knowledge Python, timeseries processing
Work distribution 100% programming
Period Summer semester 2019
First meeting Seminar introduction on

24. Apr. 2019, 17:00-18:30 at Henkestr. 91, Haus 7, 1. OG, R 373.


Up-to-date literature recommendations are provided during the lectures.


Final presentation and final report.


Rui Zhang

  • Job title: Researcher
  • Address:
    Henkestr. 91, Geb. 7
    91052 Erlangen
  • Phone number: +49 9131 85-23604
  • Email:

Friedrich-Alexander-Universität Erlangen-Nürnberg