Proceedings of the International Workshop on Advances in Deep Learning for Image Analysis and Computer Vision (IWADIC 2025)

<Previous Article In Volume
Next Article In Volume>

PGeoCLIP: Acceleration on Image geo-localization Using Precomputed Features

Authors
Chengwuzhou Wu1, *
1School of Big Data & Software Engineering, Chongqing University, Chongqing, China
*Corresponding author. Email: 20231458@stu.cqu.edu.cn
Corresponding Author
Chengwuzhou Wu
Available Online 24 April 2026.
DOI
10.2991/978-94-6239-648-7_27How to use a DOI?
Keywords
image geo-localization; precomputed features; training acceleration
Abstract

Image-based Geo-localization refers to predicting the geographic location from an image. A noble image-to-GPS retrieval approach GeoCLIP, demonstrated outstanding performance below distance threshold metrics, while the substantial training time and computational overhead present considerable challenges. Acceleration in image geo-localization refers to the enhancement of both training and inference efficiency for image geo-localization models. GeoCLIP’s time bottleneck lies in the feature extraction stage of its image encoder. In this paper, the author evaluated GeoCLIP’s performance using precomputed features on the test dataset, and proposed an improved model PGeoCLIP based on GeoCLIP, to mitigate the performance degradation caused by using precomputed features. PGeoCLIP removes random data augmentation in the dataloader and introduces several selective improvements: a gated MLP mechanism in the location encoder, as well as a supervised constraint in the image encoder. These modifications significantly reduce training time (e.g., -12h for training one epoch) and inference time (e.g., -8.70s for one query), while maintaining competitive accuracy on the test dataset.

Copyright
© 2026 The Author(s)
Open Access
Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

<Previous Article In Volume
Next Article In Volume>
Volume Title
Proceedings of the International Workshop on Advances in Deep Learning for Image Analysis and Computer Vision (IWADIC 2025)
Series
Advances in Computer Science Research
Publication Date
24 April 2026
ISBN
978-94-6239-648-7
ISSN
2352-538X
DOI
10.2991/978-94-6239-648-7_27How to use a DOI?
Copyright
© 2026 The Author(s)
Open Access
Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Cite this article

risenwbib
TY  - CONF
AU  - Chengwuzhou Wu
PY  - 2026
DA  - 2026/04/24
TI  - PGeoCLIP: Acceleration on Image geo-localization Using Precomputed Features
BT  - Proceedings of the International Workshop on Advances in Deep Learning for Image Analysis and Computer Vision (IWADIC 2025)
PB  - Atlantis Press
SP  - 242
EP  - 250
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6239-648-7_27
DO  - 10.2991/978-94-6239-648-7_27
ID  - Wu2026
ER  -