Channel-Spatial Mutual Attention Network for 360 degrees Salient Object Detection - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Channel-Spatial Mutual Attention Network for 360 degrees Salient Object Detection

Résumé

In this work, we conduct 360 degrees panoramic salient object detection by taking advantage of both the global and local visual cues of 360 degrees images, with a novel channel-spatial mutual attention network (CSMA-Net). The key component of the CSMA-Net is the proposed CSMA module, which cascades channel-/spatial-weighting-based mutual attentions. The objective of our CSMA module is to reline and fuse the bottleneck features from two separate encoders with different planar representations of 360 degrees panorama as inputs, i.e., equirectangular image and cube map. Our CSMA-Net outperforms 10 state-of-the-art segmentation methods based on the proposed 360 degrees SOD benchmark where multiple fine-tuning and testing strategies are applied to the widely-used 360 degrees datasets. Extensive experimental results illustrate the effectiveness and robustness of the proposed CSMA-Net(1).
Fichier non déposé

Dates et versions

hal-04010888 , version 1 (02-03-2023)

Identifiants

Citer

Yi Zhang, Wassim Hamidouche, Olivier Déforges. Channel-Spatial Mutual Attention Network for 360 degrees Salient Object Detection. 26th International Conference on Pattern Recognition / 8th International Workshop on Image Mining - Theory and Applications (IMTA), Aug 2022, Montreal, Canada. ⟨10.1109/ICPR56361.2022.9956354⟩. ⟨hal-04010888⟩
20 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More