Students from Saratov State University of Geology and Technology have developed an intelligent system for searching for objects in photographs.
On August 25, the Siberian State University of Geosystems and Technology successfully defended another project under the federal program "Personnel for UAS." A team of students presented and defended a solution to a complex engineering problem posed by VectorAero LLC.
The students were tasked with developing a system for searching for objects in a large image database using an image without navigational information in its metadata as a query. The developed software enables efficient matching and finding of desired objects, solving the problem of working with unlabeled or anonymous images.
The key advantages of the solution were its high efficiency and adaptability. The system demonstrated 92% object identification accuracy, processes each image in less than one second, and has a false alarm rate of only 8%, fully meeting the technical specifications. Particular attention was paid to the program's ability to operate under limited computing resources, including on portable devices and in offline mode. For user convenience, the development is equipped with an intuitive graphical interface.
The implementation of this development opens up new possibilities for analyzing large amounts of geospatial data, which is particularly relevant for mapping, territorial monitoring, and security.
