All You Need to Know About Training Image Retrieval Models

Image retrieval is the task of finding images in a database that are most similar to a given query image. The performance of an image retrieval pipeline depends on many training-time factors, including the embedding model architecture, loss function, data sampler, mining function, learning rate(s), and batch size. In this work, we run tens of thousands of training runs to understand the effect each of these factors has on retrieval accuracy. We also discover best practices that hold across multiple datasets. The code is available atthis https URL

@article{berton2025_2503.13045,
  title={ All You Need to Know About Training Image Retrieval Models },
  author={ Gabriele Berton and Kevin Musgrave and Carlo Masone },
  journal={arXiv preprint arXiv:2503.13045},
  year={ 2025 }
}