Extremely compact video representation for efficient near-duplicates detection

Open Webpage

Katarzyna Fojcik, Piotr Syga, Marek Klonowski

Published: 2025, Last Modified: 05 Nov 2025Pattern Recognit. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Downscaling pre-selected video frames creates a sequence with a distinctive pattern.•Proposed video descriptor is robust to a wide range of modifications.•Our approach enables efficient video storage and copy detection without accuracy loss.•A diverse and dynamic video dataset was created based on movie trailers.•Created dataset shows good generalization ability for video copy detection problem.