Development of the Forest Fire Satellite Detection Information System

This study describes the process of developing the forest fire satellite detection information system. In the work the impact of 2019-2020 forest fires on the economies of Russia and Australia is considered, as they were affected by massive forest fires; a review of existing forest recognition algorithms is carried out, requirements for the forest fire satellite detection information system are desribed, the business model with payback period calculation is described, using business model design template by Osterwalder and Pigneur, a Use-Case diagram for the users of the system is built and its elements are described. In addition, the main forest fire recognition algorithms (random forest, support vector machine, convolutional neural network) are analysed and their advantages and disadvantages are highlighted. After that, the convolutional neural network is designed and its structure is described, using Keras with Tensorflow frameworks, the system functions are described, using activity and sequence diagrams, and user interfaces are designed. Then, the neural network model for the forest fire satellite detection and a web application interface are built and tested, using black and white box testing. In the work a programming language Python 3.9.2 and Keras with Tensorflow library are used to build and describe the structure of the convolutional neural network model; a document markup language HTML version 5.2, a style sheet language CSS, a programming language Javascript version ECMAScript 5 are used to manage the objects of the HTML document.