Venue: Vision of Art (VISART) Workshop at the European Conference on Computer Vision (ECCV 2024)
Authors: M. Apellaniz Portos, R. Labadie-Tamayo, C. Stemmler, E. Feyersinger, A. Babic, F. Bruckner, V. Öhner, M. Zeppelzauer
This workshop paper applies weakly supervised learning to detect and analyze hybrid compositions in animation film — scenes that mix different animation techniques or combine animated and non-animated elements. Working with weak supervision keeps the approach practical when fine-grained annotations are scarce, which is typical for film archives.
The paper connects to the AniVision project on the computational analysis of animation styles, and was presented at a computer-vision-for-art workshop at ECCV.
Read the full paper: https://arxiv.org/abs/2410.04789