Document type
Bachelor thesisPublication date
Publication license
Please use this identifier to cite or link to this item: https://hdl.handle.net/2445/197380
Segmentation and classification of animal behaviour from acceleration data
Journal Title
Authors
Director/Tutor
Additional contribution
Journal ISSN
Volume Title
Abstract
[en] The study of animal behaviour is, even today, an unknown field due to the difficulty involved. Most of the time, it is unfeasible to be present to observe and analyse animal behaviour in a situation of freedom, and other study methods such as laboratory study condition the behaviour of the animal and do not allow us to study it in depth.
It has been shown that by analysing time series of the acceleration of the animal this problem can be solved, as it provides very detailed information about the movement of the animal with a high resolution over time, allowing to determine with great precision what the animal was doing at a specific time, without altering its behaviour or the need for human presence.
This work studies a new algorithm for segmenting and classifying animal acceleration data into different behaviours using tri-axial acceleration data (for each Cartesian axis), recorded using an accelerometer and placed in the red-billed tropicbird (Phaethon aethereus). This seabird lives in Cape Verde and is distinguished by flying long distances over the sea. The algorithms explained below are divided into three major blocks: segmentation, to be able to extract different behaviours from the data; grouping, to be able to cluster similar behaviours; and classification, using a recurrent network neuronal (RNN) to be able to classify previously untreated behaviours into one of the groups we found above.
Description
Treballs Finals de Grau d'Enginyeria Informàtica, Facultat de Matemàtiques, Universitat de Barcelona, Any: 2022, Director: Ignasi Cos Aguilera
Related resource
Citation
Citation
LENDÍNEZ PADILLA, Alejandro. Segmentation and classification of animal behaviour from acceleration data. [consulted: 11 of October of 2026]. Available at: https://hdl.handle.net/2445/197380