Automatic Detection of Environmental Changes Through ViT-Based Analysis of Satellite Images

Abstract

The invention relates to a system that detects environmental changes by automatically analyzing environmental changes from large satellite image data. The system includes satellite image input (1), which inputs high-resolution satellite images taken at different times into the system; image preprocessing and difference analysis (2), which normalizes time-series satellite images and performs initial difference detection; viT model (3), which learns changes by analyzing each part in the image obtained by image preprocessing and difference analysis (2) and using attention mechanisms; and an interface (4) which visualizes the environmental changes detected by the viT model (3) and generates reports and maps.

Description

Automatic Detection of Environmental Changes Through ViT-Based Analysis of Satellite Images Technical Area The invention enables the automated analysis of environmental changes from large satellite image data. It is related to a system that detects changes. State of the Art Today, the analysis of environmental changes is mostly done manually or with limited automation. It is based on traditional algorithms. Manual methods process large satellite image datasets. Traditional algorithms, on the other hand, are both time-consuming and prone to error when it comes to analysis. You may miss changes at low resolution or changes in large-scale areas. They may not be able to map accurately. Furthermore, current methods often fail to accurately represent changes. from the ability to evaluate its temporal dimension and its effects at different scales together It is lacking. Purpose of the Invention The invention was created by drawing inspiration from existing situations and overcoming the aforementioned drawbacks. It aims to solve the problem. The main goal of the invention is to automatically detect environmental changes through ViT-based analysis of satellite images. The aim is to ensure that it is detected. The invention uses the Vision Transformer (ViT) model to capture both small and large objects in satellite imagery. The model aims to accurately detect large-scale environmental changes. It analyzes time-series-based differences through multi-scale attention mechanisms and It maps environmental changes. This speeds up data analysis processes and decision-making. By providing accurate information to data collection mechanisms, significant contributions are made to the field of environmental management. This will be ensured. The invention uses satellite imagery as input to create the system. The inputs are image preprocessed and... It is normalized using difference analysis, and the difference is detected at the initial level. In the invention, the ViT model analyzes each part in the image and detects changes using attention mechanisms. It learns by analyzing image segments in the ViT model, thus learning about changes. is carried out. The invention visualizes the detected changes through an interface, generating reports and maps. 35 are created. Figures that will help understand the invention. Figure 1 shows a general representation of the system that is the subject of the invention. Description of 40 Part References 1. Satellite image input 2. Image preprocessing and difference analysis 3. ViT model 4. Interface 45 2 Detailed Description of the Invention In this detailed explanation, the preferred configurations of the system in question are described only in relation to the subject matter. It is explained in order to facilitate a better understanding. The invention enables the automated analysis of environmental changes from large satellite image data. It is a system that detects changes. The system; High-resolution satellite images taken at different times have been entered into the system. satellite image input (1), Normalizing time series satellite images and detecting differential determinants at an initial level. image preprocessing and difference analysis (2), By analyzing each part of the image obtained with image preprocessing and difference analysis (2) viT model (3), which learns changes through attention mechanisms, By visualizing the environmental changes detected by the ViT model (3), reports and interface that creates maps (4) It includes. In the system, satellite image input (1) is first given to the system and these images are then processed as image pre- Normalized by processing and difference analysis (2) step and the basic differences are determined. Then ViT With model (3), each part of the images is analyzed and attention mechanisms are used Changes are detected with high accuracy thanks to multi-scale feature extraction. The changes are analyzed on both small and large scales. Detected by the interface (4) Environmental changes are visualized and presented to decision-makers. 35 40

Claims

1. Environmental data by automatically analyzing environmental changes from large satellite imagery data. It is a system that detects changes, and its feature is; High-resolution satellite images taken at different times have been entered into the system. satellite image input (1), Normalizing time series satellite images and detecting differential determinants at an initial level. image preprocessing and difference analysis (2), By analyzing each part of the image obtained with image preprocessing and difference analysis (2) ViT model (3), which learns changes through attention mechanisms, By visualizing the environmental changes detected by the ViT model (3), reports and interface that creates maps (4) It includes.