Comparative Analysis of Selected Geotagging Methods Sichen Mu, Mateusz Piwowarczyk , Marcin Kutrzy´ nski , Bogdan Trawi´ nski (B ) , and Zbigniew Telec Faculty of Computer Science and Management, Wroc law University of Science and Technology, Wroclaw, Poland {mateusz.piwowarczyk,marcin.kutrzynski,bogdan.trawinski, zbigniew.telec}@pwr.edu.pl Abstract. Geotagging is a rapidly growing technology in digital pho- tography and searching for specific landmarks, and helps everyone in our daily lives. Navigation applications and travel guides put a number of geotagged photos on the maps, providing a good overview of the destina- tion. Recently, the development of photo geotagging methods has become a popular issue. Implementations of the SIFT and SURF algorithms and training of convolutional neural networks to obtain image classification for landmark images are presented in this paper. Based on the results of the classification of images and geotags of other similar images, geo- tags have been assigned to the target images. In addition, the results of image classification obtained using feature detection algorithms and neural networks were compared and analyzed. Keywords: Geotagging · Landmark recognition · SIFT algorithm · SURF algorithm · Convolutional neural network 1 Introduction Searching through a huge collection of images based only on their filenames could be very inefficient. Also other metadata assigned to pictures manually or automatically (e.g. coordinates of where photo is taken by camera build-in or connected GPS systems) could not be sufficient for many applications. Process of attaching geospatial identification metadata to media files such as images, videos, e-books and social media posts is called geotagging. Geotag is a form of geospatial metadata. Such data usually contains information about geolocation, timestamp, sometimes it also has GPS track information. Geotagging could be replaced by georeferencing and geocoding. Georeferencing is defined as specifying the geographic location of an object like image or movie. Geocoding is the pro- cess of finding a mathematical representation of geolocation. Generally, a geotag consists of latitude and longitude reference. In addition, the geotag may also contain legal place names. By searching through geotags, users can find different types of information at a particular location. For example, if you want to take c  Springer Nature Singapore Pte Ltd. 2020 P. Sitek et al. (Eds.): ACIIDS 2020, CCIS 1178, pp. 329–341, 2020. https://doi.org/10.1007/978-981-15-3380-8_29