Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{175494,
        author = {Gaurav Sonawane and Ganesh Dake and Nishant Ade},
        title = {SecureCAPTCHA: Smarter CAPTCHAs with Patch-Based Defense},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {11},
        pages = {4075-4080},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=175494},
        abstract = {As automated attacks against CAPTCHA systems become increasingly sophisticated, traditional text-based and image recognition challenges are becoming vulnerable to machine learning-based attacks. This paper proposes Secure CAPTCHA, a novel CAPTCHA system that incorporates strategically generated adversarial patches into images to create challenges that are easily solvable by humans but resistant to automated attacks. By introducing carefully crafted patch-based perturbations that exploit the fundamental vulnerabilities of neural networks, our system creates a robust defense mechanism against bot attacks while maintaining human usability. We train our system over 100 epochs and evaluate its performance against state-of-the-art deep learning-based CAPTCHA solvers, demonstrating significant improvements in security metrics compared to conventional approaches. Our experimental results show that Secure CAPTCHA reduces successful automated attacks by 91% while maintaining a human success rate of 94%. The system also demonstrates resilience against various adaptive attack strategies, including feature squeezing, adversarial training, and ensemble methods.},
        keywords = {CAPTCHA, Adversarial Noise, Deep Learning, Image Classification},
        month = {April},
        }