University of Twente Student Theses
Image Similarity Ranking Applied to the Forensic Domain
Bie, D.R. de (2020) Image Similarity Ranking Applied to the Forensic Domain.
|
PDF
2MB |
| Abstract: | This research focuses on investigating the possibilities of an efficient scalable image similarity ranking algorithm useful within the forensic domain. It should be able to handle large datasets of images and that is not bound to object categories. This algorithm will be used to rank images within a dataset based on similarity to an object of interest in an ongoing investigation. This research investigates the possibility of combining Convolutional Neural Networks together with distance metrics for image similarity ranking. Similarity performance analysis and different time analysis have been performed to look for a suitable algorithm. The results show that it is indeed possible to combine Convolutional Neural Networks with distance metrics to create a well performing image similarity ranking algorithm that is also able to perform well on data unknown to the network. In general, keeping similarity performance and time constraints into account, it would be best to use an InceptionResNetV2 neural network together with the cosine distance metric. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science MSc (60300) |
| Link to this item: | https://purl.utwente.nl/essays/85294 |
| 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