Institutional Repository of Vilnius University
Objektų sekimo įranga bepiločiams komerciniams orlaiviams pasitelkiant neuroninių tinklų objektų atpažinimą /
Abstract
dc:descriptionThis work analyzes unmanned aerial vehicles, their structure, use and the latest artificial intelligence software technologies. The peculiarities, advantages and disadvantages of these technologies are reviewed. A study is being conducted to select the cheapest commercial programmable drone on the market - ”Ryze Tello” drone. Object detection software is then created for it in order to detect and track a selected object. Object detection is performed using convolutional neural network object detection models. The aim of the study was to find out which type of models would give the best results in low-cost commercial aircraft. The study showed that due to the low-quality camera and low-power engines installed in the aircraft, the single-stage YOLOv5 model, which has become the industry standard for real-time object detection, performed worse than the slower, two-stage Faster R-CNN model. more commonly used to process individual photos. The speed provided by the YOLOv5 model is underutilized, and the speed achieved by the Faster R-CNN model was sufficient enough to perform real-time object detection task, and the additional accuracy allowed the aircraft to achieve better results. In poor conditions: poor lighting, poor camera quality, video artifacts, the two-stage Faster R-CNN used in the aircraft performed better on the object tracking task than the single-stage YOLOv5 model.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2022
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Sirutavičius, Tomas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD146242375&prefLang=en_US