Generation of Three-Dimensional Synthetic Datasets

The problem of automatic porting of virtual reality applications developed in the Unity game engine using native libraries of different virtual reality headsets to other headsets is considered. The problems arising during porting are identified and the algorithm of their solution is described. The presented solution is tested in the work on porting VR-applications developed earlier by the authors to different headsets, on the basis of the obtained results the general conclusion is made about the feasibility of using the developed tool for automatic porting of VR-applications. Problems yet to be solved are described and scalability possibilities are presented. With the demonstrated growth of VR use in education and in industry, the task of porting is quite broad, so the presented solution allows to achieve a significant effect if it is necessary to expand the range of headsets used.

Section: The Challenges Of Changing Vr Librariesmentioning

confidence: 99%

Approaches to automating VR applications porting using common techniques

Kugurakova,

Vasilov,

Khafizov

et al. 2024

Infrared video-based object recognition for agricultural research field uses

Guzaerov,

Sergunin,

Kugurakova

2024

The paper presents a study of an object recognition system based on infrared (IR) video stream using the YOLOv8 model. The aim of the work was to develop and evaluate the performance of a computer vision algorithm for analysing infrared images under various conditions including low light. While the use of RGB images can provide rich information in normal conditions, it becomes inefficient to use a visible spectrum camera in total darkness or in the presence of clutter. The study was based on an extensive dataset containing about 176,000 annotated frames from an infrared camera. The YOLOv8 model was adapted and trained to recognise 15 classes of objects, including people, animals and different types of vehicles. The results showed high performance of the model in recognising the most represented classes in the dataset, such as cars, people with F1-score up to 0.8. However, the problem of unbalanced classes was identified, leading to lower recognition accuracy for rarely seen objects. The study demonstrates the potential of using infrared cameras in computer vision tasks and opens new directions for further developments in the field of object recognition in limited visibility conditions. The results of the work may find application in a wide range of disciplines, including medicine, agriculture and environmental monitoring.

Creation of a synthetic dataset for training precise movements of robots for in various industries

Gabdullin,

Kugurakova,

Iskhakov

et al. 2024

Creating synthetic datasets for artificial intelligence training has a crucial role in modern developments. Considering the difficulties in collecting real data, which is often a costly and time-consuming process that requires significant resources and time. Synthetic data, on the other hand, allows generating large amounts of varied and controlled data that can be customized for specific training and testing needs. This makes the process of algorithm development and improvement more efficient and affordable. This paper presents a comprehensive tool for creating synthetic motion datasets based on rigging a 3D robot model. The ability to create and edit animations through the Blender interface is described. It supports a variety of well-known 3D model formats, providing flexibility in use, and includes powerful tools to achieve high-quality visual effects and realistic scenes. In addition, the tool can automatically generate a large number of robot images needed for training neural networks. By utilizing these capabilities, the tool greatly simplifies the creation of training datasets, making the process more efficient and affordable. Possible future enhancements include automation of rigging, further optimizing the functionality and usability of the tool for robotics and machine learning.