FFLDGA-Net: Image retrieval method based on Feature Fusion Learnable Descriptor Graph Attention Network

Xiaoyu Hu, Xingmei Wang, Dongmei Yang, Wei Ren, Jinli Wang, Bo Liu

Abstract: Highlights•Overcome the absence of depth information in images.•Introduce the LDGA-Net to mine hard samples and negative samples.•Combine the multi-scale route convolution module with path aggregation network.•Incorporate a soft label strategy tailored to mitigate training noise.