baseline.md

Baseline Experiment: Vehicle Detection

This experiment establishes a baseline for vehicle detection using a pre-trained YOLOv8 model.

Configuration

• Model: yolov8n.pt
• Dataset: A custom subset of a vehicle dataset (e.g., COCO vehicles, or a custom dataset).
• Epochs: 10
• Image Size: 640x640

Observations

• The model achieves a reasonable performance out-of-the-box.
• Further improvements can be made by fine-tuning on a larger, more specific dataset.
• The confusion matrix shows some misclassifications between 'car' and 'truck', which is expected.

Next Steps

• Fine-tune the model on a larger, custom vehicle dataset.
• Experiment with data augmentation techniques.
• Try a larger model (e.g., yolov8s.pt) for better accuracy.