Video World Models · 2026-10-01

GeoGAT: Bidirectional Temporal Sampling Meets Hierarchical Graph Attention for Global Video Geo-localization

Junchao Cui, Xuanzi Ma, Wenqi Shi, Hangyu Li, Biru Zhu, Chong Fu, Xiangyang Luo

arXiv:2609.39128PDF

AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.

TL;DR

GeoGAT addresses global video geo-localization with (1) bidirectional temporal sampling that extracts forward and offset-reversed frame sequences, (2) a graph-attention hierarchy classifier over a predefined geographic hierarchy graph, and (3) dual constraints to eliminate cross-hierarchy conflicts. They build and release GeoGAT10k (9,720 multi-shot edited videos) and report state-of-the-art across city/state/country/continent on CityGuessr68k and GeoGAT10k.

Why it matters

The authors argue that one-way uniform sampling and independent per-hierarchy classifiers lose sparse geographic cues and create cross-granularity conflicts, especially for complex multi-shot edited videos. GeoGAT’s bidirectional feature fusion plus structure-aware graph message passing and hierarchical consistency constraints directly target these failure modes, and the reported gains include both absolute improvements and conflict elimination across hierarchies.

Method

  • Bidirectional temporal sampling: encode a forward uniformly sampled frame set and an offset-reversed sampled set, fuse features into a dual representation.
  • GAT-based hierarchy classifier: project video features into a predefined directed geographic hierarchy graph and use GATv2 message passing to predict city/state/country/continent jointly.
  • Dual-constraint consistency: apply soft parent-child coupling during training and hard parent-child validity enforcement at inference to eliminate cross-hierarchy conflicts.
Abstract (from arXiv)

Global video geo-localization aims to infer the geographic location of a video worldwide, evaluating performance across four geographic hierarchies: city, state/province, country, and continent. Existing methods typically employ one-way uniform sampling to process video frames and train independent classifiers for each hierarchy, which leads to the loss of key geographic cues and prediction conflicts between hierarchies, especially for complex multi-shot edited videos. To address these limitations, we propose GeoGAT, which integrates bidirectional temporal sampling with graph attention networks (GATs). Specifically, GeoGAT extracts forward and offset-reversed frame sequences to construct complementary spatiotemporal features. These fused features are then fed into a predefined geographical hierarchy graph, where GATs perform structure-aware message passing, while a dual-constraint mechanism prunes predictions to eliminate cross-hierarchy conflicts. We construct GeoGAT10k, comprising 9,720 multi-shot edited videos from 166 cities worldwide, specifically to benchmark generalization ability on complex video structures. Experimental results on CityGuessr68k and GeoGAT10k demonstrate that GeoGAT eliminates hierarchical conflicts entirely and achieves state-of-the-art performance across all four geographic hierarchies. On CityGuessr68k, GeoGAT outperforms the strongest baseline, evaluated under both classification and retrieval protocols, by 2.6 percentage points at the city level. On the more challenging GeoGAT10k with multi-shot edited videos, the accuracy improvement exceeds 24 percentage points, validating strong generalization to complex real-world scenarios.

Related papers

Publication record

Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.

First public release: 2026-09-30 · Indexed by Pub-AI: 2026-10-01 UTC

preprint

GeoGAT: Bidirectional Temporal Sampling Meets Hierarchical Graph Attention for Global Video Geo-localization

Junchao Cui, Xuanzi Ma, Wenqi Shi, Hangyu Li, Biru Zhu, Chong Fu, Xiangyang Luo

Preprint posted: 2026-09-30 · Original record ↗

arxiv
2609.39128
Explicit source versions (1)
  • v1 · 2026-09-30
  1. Preprint available

    Source date: 2026-09-30 · First detected: 2026-10-25 UTC · Historical baseline

    Source evidence (1)
Abstract · arxiv

Global video geo-localization aims to infer the geographic location of a video worldwide, evaluating performance across four geographic hierarchies: city, state/province, country, and continent. Existing methods typically employ one-way uniform sampling to process video frames and train independent classifiers for each hierarchy, which leads to the loss of key geographic cues and prediction conflicts between hierarchies, especially for complex multi-shot edited videos. To address these limitations, we propose GeoGAT, which integrates bidirectional temporal sampling with graph attention networks (GATs). Specifically, GeoGAT extracts forward and offset-reversed frame sequences to construct complementary spatiotemporal features. These fused features are then fed into a predefined geographical hierarchy graph, where GATs perform structure-aware message passing, while a dual-constraint mechanism prunes predictions to eliminate cross-hierarchy conflicts. We construct GeoGAT10k, comprising 9,720 multi-shot edited videos from 166 cities worldwide, specifically to benchmark generalization ability on complex video structures. Experimental results on CityGuessr68k and GeoGAT10k demonstrate that GeoGAT eliminates hierarchical conflicts entirely and achieves state-of-the-art performance across all four geographic hierarchies. On CityGuessr68k, GeoGAT outperforms the strongest baseline, evaluated under both classification and retrieval protocols, by 2.6 percentage points at the city level. On the more challenging GeoGAT10k with multi-shot edited videos, the accuracy improvement exceeds 24 percentage points, validating strong generalization to complex real-world scenarios.