Abstract:Few-shot learning in visual classification aims to recognize previously unseen classes using only a few samples. While many methods based on class representations have achieved promising classification outcomes, most traditional class representations-based studies confine themselves to constructing a single level of class representation per support class, either from a global or local feature perspective. This limitation overlooks the complementary relationship between global information and local details in evaluating class distributions. To address this limitation, we propose a novel dual representations network (DR-Net) that integrates both global and local feature perspectives in this paper. Moreover, in DR-Net, recognizing the varying discriminative capacities of local features within their respective class, we draw inspiration from the ReliefF feature selection algorithm to measure the importance of local descriptors within the support classes, so as to construct global and local representations capable of perceiving both class-invariant features and distinctive features, which provide global and detailed guidance for evaluating complex class distribution. Finally, we propose an adaptive decision fusion strategy to seamlessly integrate global relationships and local relationships in probability space, naturally achieving the collaborative incorporation of global information and local details about special class. Comprehensive experiments on benchmark datasets show the evident superiority of our proposed DR-Net compared to existing methods. All datasets are publicly accessible. The code and related datasets can be found at https://github.com/116git/DR-Net.

Dual Global-Aware Propagation for Few-Shot Learning.

Learning to Propagate Labels: Transductive Propagation Network for Few-shot Learning

Dual-stream Multi-Modal Graph Neural Network for Few-Shot Learning

Collaborative Global-Local Networks for Fine-Grained Few-shot Learning.

Dual representations network for few-shot learning based on local descriptor importance: integrating global and local features

LGSim: local task-invariant and global task-specific similarity for few-shot classification

Adaptive Anchor Label Propagation for Transductive Few-Shot Learning

BDLA: Bi-directional local alignment for few-shot learning

A Few-Shot Fine-Grained Image Classification Method Leveraging Global and Local Structures

Adaptive multi-scale transductive information propagation for few-shot learning

Dual Adaptive Representation Alignment for Cross-domain Few-shot Learning

Relation Fusion Propagation Network for Transductive Few-Shot Learning

Subgraph Propagation Network for Few-shot Learning

Global Information Embedding Network for Few-Shot Learning

Hierarchical Knowledge Propagation and Distillation for Few-Shot Learning.

Dual Branch Multi-Level Semantic Learning for Few-Shot Segmentation

MDPN: Multilevel Distribution Propagation Network for Few-Shot Image Classification

Ensemble Transductive Propagation Network for Semi-Supervised Few-Shot Learning

Local feature graph neural network for few-shot learning

Few-shot learning with representative global prototype

Augmented Bi-path Network for Few-shot Learning