Moving object detection using artificial intelligence methods
Students Name: Filyk Ruslan Vasylovych
Qualification Level: magister
Speciality: Software Engineering
Institute: Institute of Computer Science and Information Technologies
Mode of Study: full
Academic Year: 2022-2023 н.р.
Language of Defence: англійська
Abstract: The purpose of this qualification work is the research of algorithms for recognizing moving objects for the purpose of their further tracking and research, as well as the development of appropriate software to provide visualization of the obtained research results on a video stream. The main difference of this software is the use of artificial intelligence models. The software is developed in the Python programming language using thirdparty libraries such as YOLO, OpenCV, and additional computational libraries Numpy and Pandas. To ensure the functionality of the neural network, an artificial intelligence model was used, as well as the Google Tensorflow library. The software’s main purpose is to process the stream of images in real-time, identify objects in the video and classify them. The key means of achieving the given task are the use of an artificial intelligence system, in particular convolutional neural networks for the analysis of digital images, as well as algorithms for the analysis of the received recognitions and their further use as a cumulative video history with the possibility of linking recognized objects from different images with each other. For the successful design and implementation of the software, a specification of requirements was drawn up according to standards and practices, based on the analysis of the subject area and business needs. In particular, this is effective planning, software product prototyping, visualization using UML diagrams, and direct software implementation of the product using currently existing tools. The work includes the results of the analysis of evaluations and implementation of algorithms for the recognition of moving objects, the application code, and an explanatory note, as well as accompanying documents, statements, and videos demonstrating the operation of the software solution and a presentation for defense. The total volume of work is 76 pages. Keywords: Object detection, object tracking, YOLO, OpenCV, Python, Mahalanobis distance, Kalman filter, Deep Sort.