University of Twente Student Theses
Visual Classification on HR-Crime dataset
Elskamp, David (2022) Visual Classification on HR-Crime dataset.
|
PDF
3MB |
| Abstract: | Being able to detect anomalies in surveillance camera footage is essential for saving time on the otherwise time-consuming process which requires manual human detection. Recently researchers have made public the HR- Crime dataset which is a subset of a larger UCF-Crime dataset. The HR-Crime subset is available for automatic visual analysis of anomalies and consists of human-related crime scenes. In this paper, we will use this subset to detect human-related anomalies. We will be building a feature extraction pipeline using the latest technologies. And we will be presenting an implementation using two visual-based approaches for detecting anomalies in surveillance footage. One makes predictions one the whole scene whereas the other will make predictions based on human proposals in a scene. The results of our approaches were compared to the previously published skeleton extraction- based approach. Our approach turned out to be useful but not as good as previous techniques. There could still be improvements made to better the results using other techniques and improving the HR-Crime dataset. Lastly, the extracted features will be made publicly available and an appendix will give further insight into our model’s prediction process. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/98330 |
| 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