This project aims to detect the objects present in satellite images using Convolutional Neural Networks (CNNs) incorporated with Regional Proposal Networks (RPNs). Object detection is an important task in the field of remote sensing and this project can be used to detect objects such as buildings, roads, and vegetation.
The SatCNNRPN model is designed to detect multiscale objects in satellite imagery, which is a crucial task for various applications in fields such as weather forecasting, urban planning, and natural resource management. The model incorporates Convolutional Neural Networks and Regional Proposal Networks to achieve high accuracy in object detection. The project focuses on three major datasets, the Stanford Drone Dataset, Airbus Aircraft Dataset, and Inria Aerial Imagery Dataset, which are commonly used in the field of satellite imagery.
Compared to other backbone architectures used for the same purpose, SatCNNRPN achieved a higher level of accuracy, making it a state-of-the-art model in this field. The term "state-of-the-art" refers to the most advanced and effective model currently available for detecting multiscale objects in satellite imagery. This research is a significant contribution to the field of satellite imagery and demonstrates the potential for further developments in object detection using deep learning. Overall, the SatCNNRPN model presents a valuable solution to the challenges of object detection in satellite imagery, and its applications extend to a wide range of fields that rely on satellite data analysis.
