Sebastian Maisel
Samuel Wunderlich
Anna Kleinhans
Publication
Shot Scale Classification using Machine Learning
We are happy to share that our PhD students / researchers Sebastian Maisel, Samuel Wunderlich and Anna Kleinhans presented their joint work on Shot Scale Classification with Machine Learning at the 2025 MVML in Paris, France
Abstract
Video or movie analysis often involves examining various audio-visual elements. One significant visual aspect that has attracted research attention is shot scale, which refers to the distance between the camera and the subject in a shot. The impact of shot scale on narrative and viewer emotion has been well researched. Additionally, methods for automatic shot classification have been developed. However, previous studies primarily focused on categorizing shot scales into discrete classes, which can lead to misclassifications. In this work, we propose a novel regression-based approach for shot scale prediction. We show that our method outperforms prior shot classification methods in cross-dataset generalization.
Authors: Sebastian Maisel, Samuel Wunderlich, Anna Kleinhans and Andreas Koch
Link to paper: Link