Abstract
Image retrieval methods for place recognition learn global image descriptors that are used for fetching geo-tagged images at inference time. Recent works have suggested employing weak and self-supervision for mining hard positives and hard negatives in order to improve localization accuracy and robustness to visibility changes (e.g. in illumination or view point). However, generating hard positives, which is essential for obtaining robustness, is still limited to hard-coded or global augmentations. In this work we propose an adversarial method to guide the creation of hard positives for training image retrieval networks. Our method learns local and global augmentation policies which will increase the training loss, while the image retrieval network is forced to learn more powerful features for discriminating increasingly difficult examples. This approach allows the image retrieval network to generalize beyond the hard examples presented in the data and learn features that are robust to a wide range of variations. Our method achieves state-of-the-art recalls on the Pitts250 and Tokyo 24/7 benchmarks and outperforms recent image retrieval methods on the rOxford and rParis datasets by a noticeable margin.
Original language | English |
---|---|
Title of host publication | 2022 International Joint Conference on Neural Networks, IJCNN 2022 - Proceedings |
Publisher | Institute of Electrical and Electronics Engineers Inc. |
ISBN (Electronic) | 9781728186719 |
DOIs | |
State | Published - 2022 |
Event | 2022 International Joint Conference on Neural Networks, IJCNN 2022 - Padua, Italy Duration: 18 Jul 2022 → 23 Jul 2022 |
Publication series
Name | Proceedings of the International Joint Conference on Neural Networks |
---|---|
Volume | 2022-July |
Conference
Conference | 2022 International Joint Conference on Neural Networks, IJCNN 2022 |
---|---|
Country/Territory | Italy |
City | Padua |
Period | 18/07/22 → 23/07/22 |
Bibliographical note
Publisher Copyright:© 2022 IEEE.
Funding
This work was supported by the Key-Area Research and Development Program of Guangdong Province (2020B0909050003), and Science and Technology Innovation Committee of Shenzhen (CJGJZD20200617102801005).
Funders | Funder number |
---|---|
Science and Technology Innovation Committee of Shenzhen | CJGJZD20200617102801005 |
Special Project for Research and Development in Key areas of Guangdong Province | 2020B0909050003 |
Keywords
- adversarial augmentation
- constrastive learning
- image retrieval
- place recognition