University of Twente Student Theses
Human/Animal Activity Recognition Data Analysis - Classification and Identification
Khawani, Aryan (2024) Human/Animal Activity Recognition Data Analysis - Classification and Identification.
|
PDF
1MB |
| Abstract: | With the rise of technology in the modern era, many pet owners and animal caretakers in general need effective and accurate activity detection. Additionally, people with elderly parents or grandparents at home want to know what their family member is doing, and if they are fine. Traditional methods of activity detection with cameras have their drawbacks, however, since cameras can be affected by only what is visible by the naked eye. Therefore, this study performs activity classification using Micro-Doppler Signature (MDS) generated from FMCW radars, comparing the differences between six different classification models for six different activity classes, which resulted in accuracies from 84% to 93% when trained and validated on a dataset of 886 Micro-Doppler Signature spectrogram images with a 20% validation split. The study goes further and proposes an architecture for a Siamese model for the identification of the subjects performing the action with an F1-score of 84% for similar pairs and 81% for dissimilar pairs. The study then concludes by adding how this work can be built upon in the future. This study is one of the first for animal activity classification and identification using MDS images. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 50 technical science in general, 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/100968 |
| 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