Abstract
The annual increase in the number of vehicles in Indonesia has created a growing need for accurate and efficient data collection. However, current methods for collecting vehicle data are manual, time-consuming, labourintensive, and prone to human error. To address these challenges, we have developed an automated system that utilizes Deep Learning to accurately count the number of vehicles at traffic light intersections. This system is based on a web application that uses YOLO for object detection and DeepSORT for object tracking. To assess the system’s effectiveness, we conducted tests using CCTV footage from highways, comparing the system’s predicted vehicle counts against actual counts. The system demonstrated robust performance, achieving a Mean Square Error (MSE) of 18.59, indicating its potential for practical deployment in traffic monitoring.
Concepts :
Access to Document
10.1109/comnetsat63286.2024.10862278SDGs
Citations by Year
| Year | Count |
|---|---|
| 2024 | 0 |