Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need? - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need?

Résumé

We show that the way inference is performed in few-shot segmentation tasks has a substantial effect on performances-an aspect often overlooked in the literature in favor of the meta-learning paradigm. We introduce a transductive inference for a given query image, leveraging the statistics of its unlabeled pixels, by optimizing a new loss containing three complementary terms: i) the crossentropy on the labeled support pixels; ii) the Shannon entropy of the posteriors on the unlabeled query-image pixels; and iii) a global KL-divergence regularizer based on the proportion of the predicted foreground. As our inference uses a simple linear classifier of the extracted features, its computational load is comparable to inductive inference and can be used on top of any base training. Foregoing episodic training and using only standard cross-entropy training on the base classes, our inference yields competitive performances on standard benchmarks in the 1-shot scenarios. As the number of available shots increases, the gap in performances widens: on PASCAL-5 i , our method brings about 5% and 6% improvements over the state-ofthe-art, in the 5-and 10-shot scenarios, respectively. Furthermore, we introduce a new setting that includes domain shifts, where the base and novel classes are drawn from different datasets. Our method achieves the best performances in this more realistic setting. Our code is freely available online: https://github.com/mboudiaf/ RePRI-for-Few-Shot-Segmentation.
Fichier principal
Vignette du fichier
Few_shot_segmentation.pdf (11.57 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03979797 , version 1 (09-02-2023)
hal-03979797 , version 2 (22-06-2023)

Identifiants

Citer

Malik Boudiaf, Hoel Kervadec, Ziko Imtiaz Masud, Pablo Piantanida, Ismail Ben Ayed, et al.. Few-Shot Segmentation Without Meta-Learning: A Good Transductive Inference Is All You Need?. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2021), Jun 2021, Nashville, TN, United States. pp.13974-13983, ⟨10.1109/CVPR46437.2021.01376⟩. ⟨hal-03979797v2⟩
25 Consultations
25 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More