| dc.description.abstract |
This research discusses the evaluation and implementation of the YOLOv4 model in detecting and tracking vehicle types in the context of road traffic. To address the research questions, the study examined the model's performance across various aspects. The results indicate that the YOLOv4 model achieved a Mean Average Precision (mAP) of 77.88% on the training dataset after 7000 iterations. The model exhibits a commendable ability to detect different vehicle types within images, with varying accuracy rates across distinct classes. The developed application within this study can record detection data for every frame within a video sequence, providing crucial information for analyzing vehicle density on roads. Despite its relatively high accuracy level, errors persist in object detection and labeling. In conclusion, this research offers insights into the capabilities and potential of the YOLOv4 model in addressing challenges related to vehicle detection in road traffic, while also identifying areas that warrant further improvement.
Keyword : YOLOv4, vehicle detection, tracking, traffic, accuracy.
| en_US |
