A team of Stanford University graduate students has developed two impressive geolocation recognition applications: PIGEON and PIGEOTTO. PIGEON uses Google Street View images, trained with OpenAI's CLIP neural network and GeoGuessr game data set, to predict image shooting locations with high accuracy. Its country prediction accuracy is as high as 92%, and the location is accurate to 25 kilometers in 40% of cases. Within. PIGEOTTO is trained on 4 million photos from Flickr and Wikipedia to realize the function of identifying locations from a single image. These two applications demonstrate the huge potential of artificial intelligence in the fields of image recognition and geolocation, providing new technical means for geographic information and image analysis.
The PIGEON and PIGEOTTO applications developed by graduate students at Stanford University achieve high-precision identification of geographical location through clever use of machine learning technology. This is not only of great significance in academic research, but also provides new possibilities for future geographical information applications and image analysis. It is worth looking forward to its further development and application.