Portfolio

Overview:

This project focuses on car segmentation and tracking using the YOLOv8l model and the Deep Sort algorithm. It processes images to segment and count vehicles, using a dataset of 4680 images. The system is designed to track cars across different classes in real-time, making it ideal for traffic analysis applications.

Key Features:

1. YOLOv8l for Car Segmentation:

The system uses YOLOv8l to detect and segment cars with high precision, identifying vehicles across different classes in real-time.

2. Deep Sort for Tracking:

By integrating the Deep Sort algorithm, the system tracks vehicles across frames, assigning unique IDs for accurate counting and analysis.

3. Diverse Dataset: Trained on 4680 images spanning 10 vehicle classes, the system is equipped to handle various types of vehicles and traffic scenarios.

4. High Performance: The system achieved an impressive mAP of 0.956, with precision at 0.905 and recall at 0.931, ensuring reliability in real-world traffic analysis.

Category:

Tags:

Links

• GitHub
• YouTube