Occupancy monitoring system: using object detection
Kondrateva, Arina (2022)
Kondrateva, Arina
2022
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2022053113505
https://urn.fi/URN:NBN:fi:amk-2022053113505
Tiivistelmä
This thesis is based on creating an algorithm for counting visitors to a public place using an object detection, as well as conducting related research. The basic motivation for the thesis is medical reasons, namely an algorithm capable of reducing the spread of respiratory viral diseases indoor. During a pandemic, there is a high risk of contracting respiratory viral diseases. Rooms that are crowded with people present a particularly high risk. In such places it is impossible to keep enough distance to be considered safe.Thus control of the occupancy of public places can help in the prevention of illnesses.
This goal was achieved by controlling the allowable number of people in a certain area. Thus, the thesis presents the work done on the creation of an algorithm that reads video from cameras. The counting of the number of people was carried out using neural networks that allow to detect a person in the frame for further tracking of movement. When creating the algorithm, an already trained YOLOv3 neural network was applied. Also, all the necessary research on the topic has been done and can be found in the thesis.
The results obtained during the research and development of the program are satisfactory. They demonstrate the ability to track a person's movement by using cameras and neural networks. As well as the possibility of further analysis for counting people. The thesis presents the tests carried out on the algorithm as well as the analysis of the results
This goal was achieved by controlling the allowable number of people in a certain area. Thus, the thesis presents the work done on the creation of an algorithm that reads video from cameras. The counting of the number of people was carried out using neural networks that allow to detect a person in the frame for further tracking of movement. When creating the algorithm, an already trained YOLOv3 neural network was applied. Also, all the necessary research on the topic has been done and can be found in the thesis.
The results obtained during the research and development of the program are satisfactory. They demonstrate the ability to track a person's movement by using cameras and neural networks. As well as the possibility of further analysis for counting people. The thesis presents the tests carried out on the algorithm as well as the analysis of the results