🌐 Google Drive: Vehicle License Plate Detector

The Vehicle License Plate Detector project aims to develop a system capable of detecting and recognizing license plates on vehicles. The project utilizes a database containing 800 training data and 100 test data. To achieve license plate recognition, the project employs TrOCR (Transformer Optical Character Recognition), a technology that uses a Transformer-based model for text recognition tasks. Through the training process, the system learns to identify and extract characters and numbers from license plates. The system achieves outstanding accuracy, with a model accuracy rate of 98.1%.

Model Architecture

This image provides an overview of the Transformer OCR (TrOCR) model architecture.

Image Augmentation

Some data augmentation techniques used include:

1. Light adjustment
2. Contrast adjustment
3. Saturation adjustment
4. Rotation

Prediction

The image below represents the model's prediction results.

Based on the test data used, an accuracy of 98.1% was achieved.