Skip to content

About

Focused on multi-digit sequence recognition from vehicle plate images using deep recurrent neural networks and image-to-sequence learning.

Topics

Resources

Stars

0 stars

Watchers

1 watching

Forks

Repository files navigation

Handwritten Multi-Digit Recognition using Deep RNNs

This project focuses on detecting and recognizing multi-digit numbers from vehicle plate images using deep learning techniques. The model utilizes Recurrent Neural Networks (RNN) to learn digit sequences from image data, based on the CAR-A and CAR-B datasets (download dataset).

Features

  • Processes vehicle plate images and extracts digit labels.
  • Uses TensorFlow/Keras for model training.
  • Supports both training and testing datasets.
  • Designed for multi-digit number recognition tasks.

Tech Stack

  • Python 3.x
  • TensorFlow / Keras
  • NumPy, Pandas, Matplotlib, PIL

Dataset Structure

ORAND-CAR-2014/
 ├── CAR-A/
 │   ├── a_train_images/
 │   ├── a_test_images/
 │   ├── a_train_gt.txt
 │   └── a_test_gt.txt
 └── CAR-B/
     ├── b_train_images/
     ├── b_test_images/
     ├── b_train_gt.txt
     └── b_test_gt.txt

How to Run

pip install -r requirements.txt
jupyter notebook checked.ipynb

Results

The model learns to recognize multi-digit sequences with high accuracy, demonstrating the effectiveness of RNNs in sequential image-to-text tasks.

License

This project is open-source under the MIT License.

About

Focused on multi-digit sequence recognition from vehicle plate images using deep recurrent neural networks and image-to-sequence learning.

Topics

Resources

Stars

0 stars

Watchers

1 watching

Forks

Used by

Contributors

Languages