Image-Based Captcha Recognition Using Deep Learning Models

ABSTRACT:

This project focuses on the development of an image-based CAPTCHA recognition system using deep learning models to detect and classify CAPTCHA characters. The system employs Convolutional Neural Networks (CNN) and MobileNet architectures, both of which are efficient in handling image classification tasks. The dataset for the project is sourced from Kaggle, containing various CAPTCHA images that mimic real-world scenarios. The CNN model is leveraged for feature extraction and character recognition, while MobileNet, known for its lightweight structure and high accuracy, is used for enhancing the model’s performance, particularly on mobile devices. The backend of the system is developed using Python, while the frontend is created using HTML, CSS, and JavaScript to provide an interactive interface for uploading CAPTCHA images. Once an image is uploaded, the model will accurately detect the individual characters in the CAPTCHA. This project demonstrates a powerful and efficient approach to CAPTCHA recognition, suitable for automation tasks.

Keywords: CAPTCHA recognition, deep learning, CNN, MobileNet, image classification, Python, HTML, CSS, JavaScript, character detection.