Dataset
The project uses a public collection of vehicle images with license-plate annotations stored in Pascal VOC XML files.
YOLOv11 · Computer Vision · Detection
Car Plate Detection with YOLOv11
A license-plate detection project that converts Pascal VOC annotations to YOLO format, trains a YOLOv11 detector and visualizes plate predictions on unseen vehicle images.
Detection preview
License plate detection view
plate0.96
Toggle the overlay to view the detected license-plate region on the vehicle image.
Project details
Training a dedicated plate detector
Convert annotations
XML bounding boxes are converted into normalized YOLO coordinates and assigned to the plate class, producing the label files required by YOLOv11.
Train and evaluate
The prepared data is organized into train and validation sets. Training monitors box loss, classification loss, precision, recall and mean Average Precision.
Visualize predictions
The trained model is tested on unseen validation images and the detected plate regions are drawn directly over the cars for a clear qualitative check.
View full detection sample ↗Future improvements
Natural next steps
OCR to read detected plate textReal-time video detectionCompare YOLOv11 model sizesHyperparameter fine-tuning