LIU, Lihao; PROST, Jean; ZHU, Lei; PAPADAKIS, Nicolas; LIÒ, Pietro; SCHÖNLIEB, Carola-Bibiane; AVILES-RIVERO, Angelica I

Métadonnées

Afficher la notice complète

Licence d’utilisation du document

LIU, Lihao
Department of Applied Mathematics and Theoretical Physics [Cambridge] [DAMTP]

PROST, Jean
Institut de Mathématiques de Bordeaux [IMB]

ZHU, Lei
Electronic and Computer Engineering Department [Hong Kong] [ECE]

Langue

Communication dans un congrès

Ce document a été publié dans

IEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR'23), 2023-06-18, Vancouver. 2022-11-13

Résumé en anglais

Shadows in videos are difficult to detect because of the large shadow deformation between frames. In this work, we argue that accounting for the shadow deformation is essential when designing a video shadow detection method. To this end, we introduce the shadow deformation attention trajectory (SODA), a new type of video self-attention module, specially designed to handle the large shadow deformations in videos. Moreover, we present a shadow contrastive learning mechanism (SCOTCH) which aims at guiding the network to learn a high-level representation of shadows, unified across different videos. We demonstrate empirically the effectiveness of our two contributions in an ablation study. Furthermore, we show that SCOTCH and SODA significantly outperforms existing techniques for video shadow detection. Code will be available upon the acceptance of this work.< Réduire

URI

https://oskar-bordeaux.fr/handle/20.500.12278/190684

Projet Européen

Nonlocal Methods for Arbitrary Data Sources

Origine

Importé de hal

Unités de recherche

Institut de Mathématiques de Bordeaux (IMB) - UMR 5251