Details
| Original language | English |
|---|---|
| Title of host publication | 2022 IEEE Winter Conference on Applications of Computer Vision |
| Subtitle of host publication | WACV 2022 |
| Pages | 1474-1484 |
| Number of pages | 11 |
| ISBN (electronic) | 978-1-6654-0915-5 |
| Publication status | Published - 2022 |
Abstract
Keywords
- cs.CV, Evaluation and Comparison of Vision Algorithms, Large-scale Vision Applications Datasets
ASJC Scopus subject areas
- Computer Science(all)
- Computer Vision and Pattern Recognition
- Computer Science(all)
- Computer Science Applications
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
2022 IEEE Winter Conference on Applications of Computer Vision: WACV 2022. 2022. p. 1474-1484.
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Interpretable Semantic Photo Geolocation
AU - Theiner, Jonas
AU - Müller-Budack, Eric
AU - Ewerth, Ralph
N1 - Funding Information: This project has partially received funding from the German Research Foundation (DFG: Deutsche Forschungsge-meinschaft, project number: 442397862).
PY - 2022
Y1 - 2022
N2 - Planet-scale photo geolocalization is the complex task of estimating the location depicted in an image solely based on its visual content. Due to the success of convolutional neural networks (CNNs), current approaches achieve super-human performance. However, previous work has exclusively focused on optimizing geolocalization accuracy. Due to the black-box property of deep learning systems, their predictions are difficult to validate for humans. State-of-the-art methods treat the task as a classification problem, where the choice of the classes, that is the partitioning of the world map, is crucial for the performance. In this paper, we present two contributions to improve the interpretability of a geolocalization model: (1) We propose a novel semantic partitioning method which intuitively leads to an improved understanding of the predictions, while achieving state-of-the-art results for geolocational accuracy on benchmark test sets; (2) We introduce a metric to assess the importance of semantic visual concepts for a certain prediction to provide additional interpretable information, which allows for a large-scale analysis of already trained models. Source code and dataset are publicly available.
AB - Planet-scale photo geolocalization is the complex task of estimating the location depicted in an image solely based on its visual content. Due to the success of convolutional neural networks (CNNs), current approaches achieve super-human performance. However, previous work has exclusively focused on optimizing geolocalization accuracy. Due to the black-box property of deep learning systems, their predictions are difficult to validate for humans. State-of-the-art methods treat the task as a classification problem, where the choice of the classes, that is the partitioning of the world map, is crucial for the performance. In this paper, we present two contributions to improve the interpretability of a geolocalization model: (1) We propose a novel semantic partitioning method which intuitively leads to an improved understanding of the predictions, while achieving state-of-the-art results for geolocational accuracy on benchmark test sets; (2) We introduce a metric to assess the importance of semantic visual concepts for a certain prediction to provide additional interpretable information, which allows for a large-scale analysis of already trained models. Source code and dataset are publicly available.
KW - cs.CV
KW - Evaluation and Comparison of Vision Algorithms
KW - Large-scale Vision Applications Datasets
UR - http://www.scopus.com/inward/record.url?scp=85126088869&partnerID=8YFLogxK
U2 - 10.48550/arXiv.2104.14995
DO - 10.48550/arXiv.2104.14995
M3 - Conference contribution
SN - 978-1-6654-0916-2
SP - 1474
EP - 1484
BT - 2022 IEEE Winter Conference on Applications of Computer Vision
ER -