HSE: Hybrid Species Embedding for Deep Metric Learning

Bailin Yang, Haoqiang Sun, Frederick W. B. Li, Zheng Chen, Jianlu Cai, Chao Song

ICCV 2023 pp. 11047-11057

doi:10.1109/ICCV51070.2023.01014 /iccv/2023/yang2023iccv-hse/

Abstract

Deep metric learning is crucial for finding an embedding function that can generalize to training and testing data, including unknown test classes. However, limited training samples restrict the model's generalization to downstream tasks. While adding new training samples is a promising solution, determining their labels remains a significant challenge. Here, we introduce Hybrid Species Embedding (HSE), which employs mixed sample data augmentations to generate hybrid species and provide additional training signals. We demonstrate that HSE outperforms multiple state-of-the-art methods in improving the metric Recall@K on the CUB-200 , CAR-196 and SOP datasets, thus offering a novel solution to deep metric learning's limitations.

PDF ICCV Semantic Scholar

Cite

Text

Yang et al. "HSE: Hybrid Species Embedding for Deep Metric Learning." International Conference on Computer Vision, 2023. doi:10.1109/ICCV51070.2023.01014

Markdown

[Yang et al. "HSE: Hybrid Species Embedding for Deep Metric Learning." International Conference on Computer Vision, 2023.](https://mlanthology.org/iccv/2023/yang2023iccv-hse/) doi:10.1109/ICCV51070.2023.01014

BibTeX

@inproceedings{yang2023iccv-hse,
  title     = {{HSE: Hybrid Species Embedding for Deep Metric Learning}},
  author    = {Yang, Bailin and Sun, Haoqiang and Li, Frederick W. B. and Chen, Zheng and Cai, Jianlu and Song, Chao},
  booktitle = {International Conference on Computer Vision},
  year      = {2023},
  pages     = {11047-11057},
  doi       = {10.1109/ICCV51070.2023.01014},
  url       = {https://mlanthology.org/iccv/2023/yang2023iccv-hse/}
}