MULTI-SCALE GRIDDED GABOR ATTENTION FOR CIRRUS SEGMENTATION - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

MULTI-SCALE GRIDDED GABOR ATTENTION FOR CIRRUS SEGMENTATION

Résumé

In this paper, we address the challenge of segmenting global contaminants in large images. The precise delineation of such structures requires ample global context alongside understanding of textural patterns. CNNs specialise in the latter, though their ability to generate global features is limited. Attention measures long range dependencies in images, capturing global context, though at a large computational cost. We propose a gridded attention mechanism to address this limitation, greatly increasing efficiency by processing multiscale features into smaller tiles. We also enhance the attention mechanism for increased sensitivity to texture orientation, by measuring correlations across features dependent on different orientations, in addition to channel and positional attention. We present results on a new dataset of astronomical images, where the task is segmenting large contaminating dust clouds.
Fichier principal
Vignette du fichier
ICIP_cirrus_camera_ready.pdf (4.77 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04062643 , version 1 (07-04-2023)

Identifiants

Citer

Felix Richards, Xianghua Xie, Adeline Paiement, Elisabeth Sola, Pierre-Alain Duc. MULTI-SCALE GRIDDED GABOR ATTENTION FOR CIRRUS SEGMENTATION. IEEE International Conference on Image Processing (ICIP), Oct 2022, Bordeaux, France. ⟨10.1109/ICIP46576.2022.9898045⟩. ⟨hal-04062643⟩
20 Consultations
34 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More