puyush / captcha_recognition

Captcha Recognition

Check out the live demo: Captcha Recognition Demo

Dataset: Captcha Images Dataset

OCR Captcha Recognition is a deep learning project designed to recognize and decipher captcha images using Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). The model is built with TensorFlow and Keras, leveraging Connectionist Temporal Classification (CTC) loss for accurate sequence labeling.

Key Features:

• End-to-End OCR Model: Integrating CNNs for image feature extraction and Bidirectional LSTMs for sequence recognition, employing residual blocks strategically within the model architecture to enhance its strength and accuracy.

• CTC Loss Function: Implementing a custom CTC loss function for training the model efficiently.

• Edit Distance Monitoring: Utilizing the Edit Distance metric to evaluate model performance and track improvements during training.

• Flexible Configuration: Easily customizable architecture and parameters for diverse captcha datasets.

If you find this project useful or interesting, please consider giving it a star! ⭐️