Improving distinctiveness in video captioning with text-video similarity

Open Webpage

Vania Velda, Steve Andreas Immanuel, Willy Fitra Hendria, Cheol Jeong

Published: 01 Jan 2023, Last Modified: 29 Jun 2025Image Vis. Comput. 2023EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Incorporating text-video similarity into rewards improves sentence distinctiveness.•The distinctiveness can be improved without sacrificing accuracy.•Performance improvement can be achieved without increasing inference time.