YOLOS Small License Plate Detection
| Property | Value |
|---|---|
| Downloads | 8,924 |
| Model Type | Vision Transformer (ViT) |
| Framework | PyTorch |
| Training Dataset | 5,200 images |
| Average Precision | 49.0% |
What is yolos-small-finetuned-license-plate-detection?
This is a specialized computer vision model based on the YOLOS (You Only Look at One Sequence) architecture, specifically fine-tuned for license plate detection. Built upon the hustvl/yolos-small base model, it combines the power of Vision Transformers with DETR loss functions to achieve robust license plate detection capabilities.
Implementation Details
The model was trained on a dataset of 5,200 images and validated on 380 images. It leverages the Vision Transformer architecture, pre-trained initially on ImageNet-1k and COCO 2017, before being fine-tuned for 200 epochs on license plate recognition data.
- Features PyTorch integration with YolosFeatureExtractor and YolosForObjectDetection
- Achieves 79.2% AP at IoU=0.50
- Particularly strong performance on large objects with 82.4% AP
Core Capabilities
- Specialized license plate detection in various conditions
- High recall rate of 89% for large objects
- Efficient processing with transformer-based architecture
- Real-time object detection capabilities
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its use of Vision Transformer architecture in license plate detection, achieving competitive accuracy while maintaining implementation simplicity. The combination of DETR loss with YOLOS architecture provides an effective balance of performance and efficiency.
Q: What are the recommended use cases?
The model is ideal for automated license plate detection systems, parking management solutions, and traffic monitoring applications. It performs best with clear, front-facing vehicle images and shows particularly strong results with large-sized license plates in the field of view.
