YOLOv11 License Plate Detection
morsetechlab/yolov11-license-plate-detection
published May 2025 · updated May 2026
YOLOv11 License Plate Detection is a detection model fine-tuned from YOLOv11 for identifying license plates in images, trained on a 10,125-image Roboflow dataset.
specs
| Task | Object Detection (License Plate Detection) |
| Architecture | YOLOv11 (n/s/m/l/x) |
| License | AGPL-3.0 |
| Input Size | 640x640 pixels |
| Training Dataset | 10,125 images (License Plate Recognition Dataset, Roboflow Universe) |
| Training Epochs | 300 |
about this model
YOLOv11-License-Plate-Detection is a detection model specialized for locating license plates in images and video frames. It is a fine-tuned variant of Ultralytics YOLOv11 (available in sizes n, s, m, l, x) trained for 300 epochs at 640×640 resolution on a public dataset of 10,125 automotive license plate images from Roboflow Universe.
Performance
Evaluation metrics for the largest variant (YOLOv11x) are shown below. These figures are likely overestimated due to train/test contamination in the upstream dataset (see warning below).
| Metric | Value |
|---|---|
| Precision | 0.9893 |
| Recall | 0.9508 |
| mAP@50 | 0.9813 |
| mAP@50-95 | 0.7260 |
Dataset Contamination
The upstream dataset exhibits train/test leakage: the same source images appear in both splits with minor augmentations. This means reported metrics do not reflect true generalization. A clean re-split with perceptual-hash deduplication and a retrained v2 with honest numbers is planned. Validate the model on your own held-out data before production use.
Known Limitations
- Fixed 640×640 inference can miss small or distant plates in high-resolution inputs (e.g., 1200×2400). The model author recommends an inference size of 1280–1600 or using tiling via SAHI.
- The model is trained primarily on automotive license plates; performance on motorcycles, non-Latin scripts, or unusual formats is not guaranteed.
The model is released under the AGPL-3.0 license.
best for
- ·Smart Parking Systems
- ·Tollgate / Access Control Automation
- ·Traffic Surveillance & Enforcement
- ·ALPR with OCR Integration
FAQ
License plate detection in smart parking, tollgate, traffic surveillance, and ALPR systems.
AGPL-3.0.
Images at 640x640 resolution, compatible with YOLOv5 format (images plus label txt).
Use the gigarouter OpenAI-compatible endpoint with an API key.
Yes, the upstream dataset has train/test leakage; metrics are overestimated. Validate on your own data before production use.
We're benchmarking and onboarding YOLOv11 License Plate Detection as a hosted, OpenAI-compatible API. Sign in for free credit and be ready when it lands, or tell us you want it and we'll prioritize it.