Multimodal large language models (MLLMs) have advanced image geolocalization mainly by improving how they reason about geographic cues. How that reasoning isdecoded into coordinates, however, has lagged behind.
arXiv:2606. 25128v1 Announce Type: cross Abstract: Volume and quality of datasets are crucial for deep learning model training, yet they are often constrained by availability and data acquisition costs.
arXiv:2504. 11171v5 Announce Type: replace-cross Abstract: We present TerraMind, the first any-to-any generative, multimodal foundation model for Earth observation (EO).
S$^3$Geo introduces a structure‑semantic synergistic learning framework for cross‑view geo‑localization, combining a Decoupled Query Pooling module to capture local structural patterns, an optimal‑transport based contrastive learning scheme for soft cross‑view correspondences, and a Semantic Knowledge Distillation strategy from a frozen CLIP teacher to embed semantic priors. The method jointly models fine‑grained structural and semantic information, enabling robust matching between images from vastly different viewpoints such as drone and satellite views. Experiments on University‑1652 and SUES‑200 datasets show that S$^3$Geo outperforms state‑of‑the‑art approaches without increasing inference complexity.
Volume and quality of datasets are crucial for deep learning model training, yet they are often constrained by availability and data acquisition costs. Synthetic data augmentation can extend existing datasets with realistic images, and the quality of these images is generally assessed through fidelity metrics such as FID, KID, IS, LPIPS and SSIM that measure structural or distributional similarity.
arXiv:2603.08491v3 Announce Type: replace Abstract: Cross-modal Geo-localization (CMGL) matches ground-level text descriptions with geo-tagged aerial imagery, which is crucial for pedestrian navigati...
Open-vocabulary remote sensing segmentation has recently emerged as a promising paradigm that enables pixel-level recognition of arbitrary categories specified by natural language, including classes unseen during training. However, geospatial domain shifts caused by heterogeneous regions, spatial resolutions, and acquisition platforms weaken visual-text matching and limit cross-dataset generalization.
The paper introduces a composition‑aware pretraining framework for geospatial foundation models that explicitly encodes fractional land‑cover mixtures as histogram targets for each satellite image cell. By using Earth Mover’s Distance to distill these composition targets into a 36.8 M‑parameter backbone, the authors demonstrate significant improvements on region‑level tasks such as zero‑shot image retrieval and scene classification, while maintaining competitive performance on fine‑grained tasks like segmentation and object detection. The method outperforms larger models (SatMAE and Prithvi‑EO‑2.0) and achieves a 55.6 % relative boost on the ForestNet‑12 dataset, evidencing the benefit of explicit composition modeling.
The paper shows that the signal‑to‑noise ratio (SNR) distribution is key to achieving fine‑grained spatial control in flow‑matching generative models. By training on high‑noise regimes rather than a centered SNR distribution, the authors raise layout adherence from 25.9 % to 80.8 % on a synthetic benchmark, approaching the 83.6 % ceiling of the underlying variational autoencoder. These insights also improve real‑world layout accuracy on COCO, boosting average precision from 1.0 % to 57.3 % without changing the model architecture.
arXiv:2610.20592v1 Announce Type: new Abstract: Existing geometry-aware Multimodal Large Language Models (MLLMs) increasingly rely on external 3D encoders to provide structural priors, but executing...
The paper introduces GeoNeXt, a framework that repurposes pretrained video generative models for geometry estimation by framing it as a next‑frame prediction task. Unlike prior methods that either train separate depth/normal models or fine‑tune image diffusion backbones, GeoNeXt jointly models images and geometric targets, leveraging the structured knowledge of video models for more data‑efficient learning. Experiments show zero‑shot monocular depth and surface normal estimation that outperforms existing generative approaches and rivals discriminative state‑of‑the‑art methods while using far less training data.
arXiv:2606.08918v2 Announce Type: replace Abstract: Worldwide image geo-localization aims to determine where on Earth a single image was captured. However, visually similar scenes may lie thousands o...