Generating Ground Truth Images Using GAN (Generative Adversarial Network)

EasyChair Preprint 8887

Abstract

Image segmentation is the classification of each pixel on the image and creating a map of all the object areas on the image. Segmentation, biomedical, environment, geography, etc. It is widely used in the analysis of images in many fields. Alongside the source images for segmentation, corresponding basic reality images must be available. But manually obtaining basic reality information takes a lot of effort and time. In this study, our aim is to develop a deep learning-based system that can generate basic reality images from satellite images. Satellite images are trained using GAN (Generative Adversarial Network) architecture. The model takes satellite image as input and generates target image (google map image) as output. The proposed system outputs the google map image corresponding to a particular satellite image and the expected google map image. Similarity can be analyzed visually.

Keyphrases: GAN, Görüntüden görüntüye dönüşüm, Pix2Pix, Segmentasyon, Uydu Görüntüsü

@booklet{EasyChair:8887,
  author    = {Elif Işılay Ünlü and Ahmet Çınar},
  title     = {Generating Ground Truth Images Using GAN (Generative Adversarial Network)},
  howpublished = {EasyChair Preprint 8887},
  year      = {EasyChair, 2022}}