Automatic Outdoor Image Geolocation with Focal Modulation Networks

We address the problem of estimating a photo’s geographical location. Success in this estimation enables many impactful applications, like facilitating Disaster Management circumstances. However, is also very challenging task. Due to complexity problem, we restrict area geolocation single city, treating as classification where districts city are classes be distinguished. In paper, exploit Focal Modulation Network that proven perform effectively and efficiently visual modeling for real-world applications. Experimental results on two diverse datasets, crawled from online sources, show effectiveness our approach. can geolocate correctly more than two-thirds test images larger dataset about one-third an experimental training ten-times smaller size.