Semantic Segmentation of Satellite Images: A Survey
Abstract
In the era of AI, Computer Vision plays an important role in various applications which are beneficial for the society. Semantic Segmentation is one such sub domain of Computer Vision which has a number of applications such as Autonomous Driving, Medical Image Diagnosis, Satellite Image Processing etc. In simple words, Semantic Segmentation is the process of assigning pixels to different classes present in a visual imagery. It is important for researchers working in this field to know about some of the widely used Semantic Segmentation models. Our main focus is on Semantic Segmentation of Satellite Images. Studying Satellite Images and using it for better understanding of our planet is possible due to advancements in Computer Vision and Deep Learning along with availability of low cost high performance GPUs. This paper provides literature survey of various Semantic Segmentation models that can be used for processing various Satellite Images. The paper also discusses about Satellite Image processing techniques, its challenges, various Satellite Images datasets and different evaluation metrics used for the purpose of evaluation of these Semantic Segmentation models.