CNN Image Retrieval Learns from BoW: Unsupervised Fine-Tuning with Hard Examples

Filip Radenovic, Giorgos Tolias, Ondrej Chum

ECCV 2016 pp. 3-20

doi:10.1007/978-3-319-46448-0_1 /eccv/2016/radenovic2016eccv-cnn/

Abstract

Convolutional Neural Networks (CNNs) achieve state-of-the-art performance in many computer vision tasks. However, this achievement is preceded by extreme manual annotation in order to perform either training from scratch or fine-tuning for the target task. In this work, we propose to fine-tune CNN for image retrieval from a large collection of unordered images in a fully automated manner. We employ state-of-the-art retrieval and Structure-from-Motion (SfM) methods to obtain 3D models, which are used to guide the selection of the training data for CNN fine-tuning. We show that both hard positive and hard negative examples enhance the final performance in particular object retrieval with compact codes.

PDF ECCV Semantic Scholar

Cite

Text

Radenovic et al. "CNN Image Retrieval Learns from BoW: Unsupervised Fine-Tuning with Hard Examples." European Conference on Computer Vision, 2016. doi:10.1007/978-3-319-46448-0_1

Markdown

[Radenovic et al. "CNN Image Retrieval Learns from BoW: Unsupervised Fine-Tuning with Hard Examples." European Conference on Computer Vision, 2016.](https://mlanthology.org/eccv/2016/radenovic2016eccv-cnn/) doi:10.1007/978-3-319-46448-0_1

BibTeX

@inproceedings{radenovic2016eccv-cnn,
  title     = {{CNN Image Retrieval Learns from BoW: Unsupervised Fine-Tuning with Hard Examples}},
  author    = {Radenovic, Filip and Tolias, Giorgos and Chum, Ondrej},
  booktitle = {European Conference on Computer Vision},
  year      = {2016},
  pages     = {3-20},
  doi       = {10.1007/978-3-319-46448-0_1},
  url       = {https://mlanthology.org/eccv/2016/radenovic2016eccv-cnn/}
}