University of Twente Student Theses


Analyzing descriptive captions for crime recognition in surveillance footage

Albertsson, Vincent Gustav (2023) Analyzing descriptive captions for crime recognition in surveillance footage.

[img] PDF
Abstract:Crime recognition in surveillance footage is a complicated and challenging task that can benefit from computer vision and machine learning techniques. With an immense amount of footage being recorded by security cameras all the time, automated video analysis techniques can help balance the workload. This paper proposes a novel approach that extracts descriptive captions from surveillance footage for crime recognition to support crime investigation and prevention. A pipeline is adapted to extract captions, process them, and train a model for crime recognition with the HR-Crime dataset. The aim is to evaluate the effectiveness of captions for crime recognition and compare the results with state-of-the-art methods. The findings contribute to the body of research on crime detection by highlighting the potential of this approach, and providing insights into its limitations and further work that can be done.
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:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page