Published Paper Details:

CAPTCHAs are widely used to differentiate between human users and automated bots, ensuring security in online interactions. However, traditional CAPTCHA systems often suffer from usability issues and vulnerabilities to evolving machine learning-based attacks. This research presents a novel machine learning-driven approach to refining CAPTCHA mechanisms, enhancing both security and user experience. Our proposed solution leverages deep learning models to analyse and improve CAPTCHA complexity, making it more resistant to automated solvers while maintaining accessibility for genuine users. By utilizing advanced image processing and adaptive challenge generation techniques, the system dynamically adjusts CAPTCHA difficulty based on real-time threat analysis, reducing friction for legitimate users while thwarting bots. Experimental results demonstrate that our approach significantly improves CAPTCHA robustness against automated attacks while ensuring a seamless experience for human users. This research contributes to the ongoing development of secure and user-friendly authentication mechanisms, bridging the gap between security and usability in modern web applications