🤖 AI Summary
This study addresses the challenge of visual geolocation in ASEAN countries, where highly similar urban landscapes hinder accurate and interpretable localization. To tackle this issue, the authors propose the first explainable, nation-level image geolocation framework covering all eleven ASEAN nations. The approach integrates CLIP zero-shot classification with LightGBM and MLP classifiers, and introduces a novel multi-dimensional interpretability pipeline combining CLIP attention rollout, YOLOv8 object detection, and Energy-Based Perturbation Games (EBPG) to uncover the key visual cues driving predictions. Experimental results demonstrate that the MLP model achieves 85.91% accuracy and F1 score on the test set while providing object-level fine-grained interpretability, revealing a notable discrepancy between frequently detected objects and regions receiving high attention weights.
📝 Abstract
Image geolocation aims to infer the geographic origin of an image from visual content alone. However, this task remains challenging in regions where countries share similar urban, roadside, architectural, and environmental characteristics. Many existing geolocation models focus on coordinate level prediction or classification performance while providing limited insight into how visual evidence contributes to location predictions. This study presents an explainable country level image geolocation pipeline for 11 ASEAN countries. First, we collected 4,850 images from GeoGuessr style sources, Google Images, and additional street level imagery. We then evaluated three approaches on this dataset: CLIP zero shot classification, a LightGBM classifier, and an MLP classifier. The MLP achieved the best test performance, attaining an accuracy and F1 score of 85.91%. For explainability, predictions generated by the MLP classifier were analyzed post hoc using CLIP attention rollout, YOLO26 object detection on the original images, and Energy Based Pointing Game (EBPG) overlap metrics. Object level analysis indicates that frequently detected objects are not necessarily associated with the highest attention density, suggesting that object frequency and attention based visual evidence capture different aspects of a scene. These results demonstrate that the proposed model can support accurate regional image geolocation while enabling object level inspection of the visual cues underlying its predictions.