# Spatiotemporal Contrastive Learning for Cross-View Video Localization in Unstructured Off-road Terrains

arXiv: https://arxiv.org/abs/2506.05250 · HTML: https://mlsanity.com/paper/2506.05250/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 22 references that are in the map.
- First posted on arXiv 5 Jun 2025.

## Builds on (most cited first, 15 of 15)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Patch-NetVLAD: Multi-Scale Fusion of Locally-Global Descriptors for Place Recognition](https://mlsanity.com/paper/2103.01486/) (2021)
- [Coming Down to Earth: Satellite-to-Street View Synthesis for Geo-Localization](https://mlsanity.com/paper/2103.06818/) (2021)
- [TartanVO: A Generalizable Learning-based VO](https://mlsanity.com/paper/2011.00359/) (2020)
- [Fine-Grained Cross-View Geo-Localization Using a Correlation-Aware Homography Estimator](https://mlsanity.com/paper/2308.16906/) (2023)
- [Boosting 3-DoF Ground-to-Satellite Camera Localization Accuracy via Geometry-Guided Cross-View Transformer](https://mlsanity.com/paper/2307.08015/) (2023)
- [Accurate 3-DoF Camera Geo-Localization via Ground-to-Satellite Image Matching](https://mlsanity.com/paper/2203.14148/) (2022)
- [Super Odometry: IMU-centric LiDAR-Visual-Inertial Estimator for Challenging Environments](https://mlsanity.com/paper/2104.14938/) (2021)
- [GAMa: Cross-view Video Geo-localization](https://mlsanity.com/paper/2207.02431/) (2022)
- [TartanDrive 2.0: More Modalities and Better Infrastructure to Further Self-Supervised Learning Research in Off-Road Driving Tasks](https://mlsanity.com/paper/2402.01913/) (2024)
- [GAReT: Cross-view Video Geolocalization with Adapters and Auto-Regressive Transformers](https://mlsanity.com/paper/2408.02840/) (2024)
- [Wide-Area Geolocalization with a Limited Field of View Camera](https://mlsanity.com/paper/2209.11854/) (2022)
- [BEVRender: Vision-based Cross-view Vehicle Registration in Off-road GNSS-denied Environment](https://mlsanity.com/paper/2405.09001/) (2024)
- [BEVLoc: Cross-View Localization and Matching via Birds-Eye-View Synthesis](https://mlsanity.com/paper/2410.06410/) (2024)

## Built on by (most cited first, 2 of 2)

- [TransLocNet: Cross-Modal Attention for Aerial-Ground Vehicle Localization with Contrastive Learning](https://mlsanity.com/paper/2512.10419/) (2025)
- [Cross-View Sequential Visual Localization with Spatio-Temporal Context Modeling for Autonomous Driving](https://mlsanity.com/paper/2608.10660/) (2026)

## Abstract (arXiv)

Robust cross-view 3-DoF localization in GPS-denied, off-road environments remains challenging due to (1) perceptual ambiguities from repetitive vegetation and unstructured terrain, and (2) seasonal shifts that significantly alter scene appearance, hindering alignment with outdated satellite imagery. To address this, we introduce MoViX, a self-supervised cross-view video localization framework that learns viewpoint- and season-invariant representations while preserving directional awareness essential for accurate localization. MoViX employs a pose-dependent positive sampling strategy to enhance directional discrimination and temporally aligned hard negative mining to discourage shortcut learning from seasonal cues. A motion-informed frame sampler selects spatially diverse frames, and a lightweight temporal aggregator emphasizes geometrically aligned observations while downweighting ambiguous ones. At inference, MoViX runs within a Monte Carlo Localization framework, using a learned cross-view matching module in place of handcrafted models. Entropy-guided temperature scaling enables robust multi-hypothesis tracking and confident convergence under visual ambiguity. We evaluate MoViX on the TartanDrive 2.0 dataset, training on under 30 minutes of data and testing over 12.29 km. Despite outdated satellite imagery, MoViX localizes within 25 meters of ground truth 93% of the time, and within 50 meters 100% of the time in unseen regions, outperforming state-of-the-art baselines without environment-specific tuning. We further demonstrate generalization on a real-world off-road dataset from a geographically distinct site with a different robot platform.