Tezislər / Theses

THE 3
rd
 INTERNATIONAL SCIENTIFIC CONFERENCES OF STUDENTS AND YOUNG RESEARCHERS 
dedicated to the 99
th
anniversary of the National Leader of Azerbaijan Heydar Aliyev
173
RASPBERRY PI BASED SYSTEM FOR VISUAL OBJECT 
DETECTION AND TRACKING 
Ramil Abbaszade 
Baku Higher Oil School 
Baku, Azerbaijan 
[e-mail] 
Supervisor: PhD, Associate Professor Leyla Muradkhanli 
Keywords: 
Python, Yolo algorithm, Raspberry Pi, OPENCV 
Detecting and tracking objects is one of the most common and difficult 
jobs that a monitoring system must do in evaluate relevant events and 
suspicious actions, as well as automatically analyze and extract video 
information. An object might be a face, a head, a human, a line of people, a 
crowd, or a product on an assembly line, according to the business 
intelligence concept.
Object detection is undergoing a fast revolution in the 
field of computer vision. Its role in both object classification and object 
localization makes it one of the most difficult problems in the field of computer 
vision. The purpose of this detection approach is to establish where objects 
are situated in a given image, a process known as object localization, and to 
which category each item belongs, a process known as object classification.
In this paper, I will provide the reader with the most common trends in 
terms of Visual Object Detection and Tracking. Object identification methods 
include rapid R-CNN, Retina-Net, and Single-Shot MultiBox Detector (SSD). 
Despite the fact that these techniques have overcome the constraints of data 
limitation and modeling in object identification, they are not capable of 
detecting objects in a single algorithm run. There are 8 most common 
algorithms for object detection. The one which I will use is called YOLO (You 
Only Look Once). This algorithm is well-known for its speed and precision. It 
has been used to identify traffic signals, pedestrians, parking meters, and 
animals in a variety of applications. 
YOLO is implemented using OpenCV deep learning library in Python. 
YOLO performs object detection as a regression problem and returns the 
class probabilities of the collected images. To detect objects in real-time, the 
YOLO adapted convolutional neural networks (CNN). To identify objects, the 
approach requires only a single forward propagation through a neural 
network, as the name implies. This indicates that the complete image is 
predicted in a single algorithm run. The CNN is used to forecast several class 
probabilities and bounding boxes at the same time. There are several 
variations of the YOLO algorithm. Tiny YOLO and YOLOv3 are two popular 
examples. 
Advantages of YOLO 

---