AE-Net: Fine-grained sketch-based image retrieval via attention-enhanced network
作者：
Highlights：
• A novel FG-SBIR model with Attention-enhanced Network (AE-Net) is established, which pays more attention to the fine-grained details of the sketches and images.
• We introduce three modules, i.e., the Residual Channel Attention module, Local Self-attention mechanism, and Spatial Sequence Transformer to mine the fine-grained details of the sketches and images in all dimensions.
• Mutual Loss is proposed to improve the traditional Triplet Loss and restrain the distance relations among the sketches/images in a single modality.
摘要
•A novel FG-SBIR model with Attention-enhanced Network (AE-Net) is established, which pays more attention to the fine-grained details of the sketches and images.•We introduce three modules, i.e., the Residual Channel Attention module, Local Self-attention mechanism, and Spatial Sequence Transformer to mine the fine-grained details of the sketches and images in all dimensions.•Mutual Loss is proposed to improve the traditional Triplet Loss and restrain the distance relations among the sketches/images in a single modality.
论文关键词：Fine-grained sketch-based image retrieval (FG-SBIR),Residual channel attention,Local self-spatial attention,Contrastive learning,Spatial sequence transformer
论文评审过程：Received 10 June 2020, Revised 21 August 2021, Accepted 30 August 2021, Available online 1 September 2021, Version of Record 17 September 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.108291
