Abstract. This paper presents a deep learning and classical computer vision framework for cross-view geolocalization using 360-degree multi-perspective view (PV) images and an offline global map. Recent studies on cross-view geolocalization typically rely on deep learning models to localize panoramic PV images by matching them with reference satellite imagery. However, such approaches face practical limitations in real-world deployments, due to their dependence on large-scale GPU resources and the need to store extensive satellite image datasets. To address these challenges, we propose BEV-LOC, a lightweight and real-time cross-view geolocalization method. BEV-LOC employs Bird’s Eye View (BEV) encoder that learns to transform 360-degree multi-PV images into a local high-definition (HD) BEV map. The localization is then performed using Intersection Over Union (IoU)-based template matching with an offline global map. Our architecture achieves real-time performance at 30 FPS without the need for high-end GPU hardware and delivers a high positioning accuracy of 1.2 meters.
J. Kwag, C. Toth, Alper Yilmaz· The International Archives o...· 0 citations
Abstract. Structure-from-Motion (SfM) pipelines rely heavily on the detection and matching of repeatable keypoints across images, yet the performance of modern learned feature extractors in challenging environments remains insufficiently understood. This paper evaluates classical and deep keypoint detectors for SfM reconstruction using winter Arctic UAV imagery, a domain characterized by low texture, repetitive patterns, and limited man-made structure. We compare three feature pipelines within a shared PyCOLMAP-based framework: SIFT with nearest-neighbor matching (SIFT+NN), SuperPoint, and DISK, along with a hybrid approach combining SuperPoint and DISK correspondences. Quantitative evaluation is conducted using standard SfM metrics, including number of observations, track length, observations per image, and reprojection error, complemented by qualitative analysis of keypoint distributions and reconstruction interpretability. Results show that SIFT+NN consistently achieves the most complete and stable reconstructions, producing the highest number of matched observations and lowest reprojection error across aggregate experiments. However, on more challenging subsets lacking clear structural features, learned methods demonstrate improved robustness, successfully reconstructing multiple views where SIFT fails. SuperPoint provides broader spatial coverage, while DISK produces denser clusters in high-confidence regions, highlighting complementary behaviors between learned approaches. Overall, the findings indicate that classical methods remain strong baselines for Arctic UAV photogrammetry under standard SfM pipelines, while learned detectors offer advantages in difficult conditions. The observed performance gap is attributed to domain mismatch and backend optimization for handcrafted features. These results suggest that domain-specific training and improved spatial feature distribution are promising directions for advancing learned keypoint methods in Arctic reconstruction tasks.
Nicholas Sansoterra, M. G. Lenzano, William J. Shuart et al.· The International Archives o...· 0 citations