DEVELOPMENT OF MULTI-CAMERA OBJECT TRACKING SYSTEM WITH DEEP LEARNING FOR SMART SURVEILLANCE
The use of advanced surveillance technology is increasingly rising in response to the growing complexity of security needs. One of the main challenges in security systems is the ability to track multiple objects across various areas simultaneously. Therefore, this final project aims to develop a...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/82422 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | The use of advanced surveillance technology is increasingly rising in response to
the growing complexity of security needs. One of the main challenges in security
systems is the ability to track multiple objects across various areas simultaneously.
Therefore, this final project aims to develop a multi-camera object tracking system
using deep learning for smart surveillance.
This final project begins with the background of the importance of smart
surveillance that can integrate multiple cameras and track numerous objects
simultaneously. The main objective of this research is to create a system capable of
overcoming the limitations of conventional surveillance systems, which often can
only monitor limited areas and a limited number of objects.
The methods used in the development of this system involve several stages,
including feature matching from various cameras, image processing for object
detection, and object tracking algorithms.
Hypothesis analysis and verification are conducted by testing the system in various
scenarios, such as distances ranging from 1 meter to 5 meters and camera angles
from 0 degrees to 90 degrees. The test results show that the LoFTR algorithm has
high accuracy in feature matching at close distances and small camera angles, but
its accuracy decreases at longer distances and larger angles. Additionally, the
system demonstrates good performance in counting the number of detected objects
and calculating the time objects are detected.
The contribution of this final project is the development of a more advanced and
effective smart surveillance application, which is not only useful for enhancing
security but also has potential applications in other fields. Thus, this final project
provides an innovative solution that can be applied in various areas to improve the
efficiency and security of surveillance. |
|---|