In this paper, we address the challenge of visual geo-localization from low-quality UAV imagery captured in real world environments. We propose a two-stage architecture, which includes Super-Resolution and visual geo-localization. We introduced novel, non-learnable Ensemble Super-Resolution (ESR) module, which first refines upscaled aerial frames, then seamlessly feeds the enhanced imagery into a visual geo-localization pipeline.
Досліджено особливості інтелектуальних компонент інтегрованих автоматизо- ваних систем управління. Розглянуто створення інтелектуальної компоненти ІАСУ для енергетичних систем. Запропоновано нейромережні методи прогнозування споживання електроенергії підприємством на основі машини геометричних перетворень. Наведено результати проведених експериментів.
In this article the methods of image superresolution based on machine learning are
investigated. The work of different groups of these methods are analyzed. Basic features of this
methods are describing. On the basis of practical experiments comparative analysis (by the
criterion PSNR) of the superresolution methods in the case of one input image from different
classes were conducted. Experimentally found that the best results are obtained in case of
using the method based on the convolutional neural network. Despite the requirement on the