YOLOv5m License Plate
keremberke/yolov5m-license-plate
published Jan 2023 · updated Jan 2023
YOLOv5m License Plate is a detection model that identifies and localizes license plates in images.
specs
| Task | Object Detection |
| Architecture | YOLOv5m (medium variant) |
| Input Size | 640x640 pixels (default) |
| [email protected] | 0.988 on validation set |
| Training Data | keremberke/license-plate-object-detection (8,820 images) |
about this model
keremberke/yolov5m-license-plate is an object detection model that localizes and classifies license plates in images using the YOLOv5m architecture.
The model was fine-tuned on the keremberke/license-plate-object-detection dataset, which contains 8,820 annotated images (6,182 training, 1,770 validation, 882 test) with a single class (license plate). The dataset is tagged for Self Driving and Automatic Number Plate Recognition (ANPR) domains.
On the validation split, the model achieves a mean Average Precision at IoU 0.5 ([email protected]) of 0.9883, indicating high detection accuracy. Because the model predicts only one class, inference requires no class-agnostic NMS and benefits from the YOLOv5m trade-off between speed and precision.
Dataset splits
| Split | Images |
|---|---|
| Train | 6,182 |
| Validation | 1,770 |
| Test | 882 |
| Total | 8,820 |
No latency or throughput benchmarks are published for this specific model. Gigarouter hosts the model as a managed, OpenAI-compatible API—users send images and receive bounding boxes and confidence scores without managing infrastructure.
best for
- ·Automatic number plate recognition (ANPR) for parking lot entry/exit systems
- ·Traffic monitoring and toll collection
- ·Vehicle access control in gated communities or secure facilities
FAQ
It detects a single class: license plates. Each detection outputs a bounding box with a confidence score.
It achieves a mean average precision ([email protected]) of 0.988 on the validation split of the training dataset.
It expects images at a default resolution of 640x640 pixels, but can be resized automatically. Input can be a URL or base64-encoded image.
Yes, it is deployed as an OpenAI-compatible API on gigarouter. Use an API key to send requests to the endpoint.
The model card does not specify a license. Please check the Hugging Face repository for the latest terms.
We're benchmarking and onboarding YOLOv5m License Plate 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.