LocDiffusion: Identifying Locations on Earth by Diffusing in the Hilbert Space

Zhangyu Wang, Jielu Zhang, Zhongliang Zhou, Qian Cao, Nemin Wu, Zeping Liu, Lan Mu, Yang Song, Yiqun Xie, Ni Lao, Gengchen Mai

Keywords: Image geo-localization, location encoding, diffusion

Abstract:

Image geolocalization is a fundamental yet challenging task, aiming at inferring the geolocation on Earth where an image is taken. Existing methods approach it either via grid-based classification or via image retrieval. The geolocalization accuracy of these methods is constrained by the choice of geographic grid cell sizes or the spatial distributions of the retrieval image/geolocation gallery, and
their performance significantly suffers when the spatial distribution of test images does not align with such choices. To address these limitations, we propose to leverage diffusion models to achieve image geolocalization with arbitrary resolutions. To avoid the problematic manifold reprojection step in diffusion, we developed a novel spherical positional encoding-decoding framework, which encodes points on a spherical surface (e.g., geolocations on Earth) into a Hilbert space of Spherical Harmonics coefficients and decodes points (geolocations) by mode-seeking. We call this type of position encoding Spherical Harmonics Dirac Delta (SHDD) Representation. We also propose a novel SirenNet-based architecture called CS-UNet to learn the conditional backward process in the latent SHDD space by minimizing a latent KL-divergence loss. We train a conditional latent diffusion model called LocDiffusion that generates geolocations under the guidance of images – to the best of our knowledge, the first generative model to address the image geolocalization problem. We evaluate our LocDiffusion model against SOTA image geolocalization baselines. LocDiffusion achieves competitive geolocalization performance and demonstrates significantly stronger generalizability to unseen geolocations.

Primary Area: applications to computer vision, audio, language, and other modalities

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 6982