This is an official implementation in PyTorch of SpatialFormer, which is accepted by AAAI-2023.

The full code will be released soon. This code is based on the implementations of tSF: Transformer-based Semantic Filter for Few-Shot Learning.

Recent Few-Shot Learning (FSL) methods put emphasis on generating a discriminative embedding features to precisely measure the similarity between support and query sets. Current CNN-based cross-attention approaches generate discriminative representations via enhancing the mutually semantic similar regions of support and query pairs. However, it suffers from two problems: CNN structure produces inaccurate attention map based on local features, and mutually similar backgrounds cause distraction. To alleviate these problems, we design a novel SpatialFormer structure to generate more accurate attention regions based on global features. Different from the traditional Transformer modeling intrinsic instancelevel similarity which causes accuracy degradation in FSL, our SpatialFormer explores the semantic-level similarity between pair inputs to boost the performance. Then we derive two specific attention modules, named SpatialFormer Semantic Attention (SFSA) and SpatialFormer Target Attention (SFTA), to enhance the target object regions while reduce the background distraction. Particularly, SFSA highlights the regions with same semantic information between pair features, and SFTA finds potential foreground object regions of novel feature that are similar to base categories. Extensive experiments show that our methods are effective and achieve new state-of-the-art results on few-shot classification benchmarks.

If you use this code for your research, please cite our paper:

@article{lai2023spatialformer,
  title={SpatialFormer: Semantic and Target Aware Attentions for Few-Shot Learning},
  author={Lai, Jinxiang and Yang, Siqian and Wu, Wenlong and Wu, Tao and Jiang, Guannan and Wang, Xi and Liu, Jun and Gao, Bin-Bin and Zhang, Wei and Xie, Yuan and Wang, Chengjie},
  journal={AAAI},
  year={2023}
}

This code is based on the implementations of tSF: Transformer-based Semantic Filter for Few-Shot Learning.