aiCAPTCHA relies on humans' innate ability to locate and identify objects compared to the challenges automated algorithms encounter doing the same. aiCAPTCHA tests contain a composite of many photographs of different types of objects. Users are asked to select the images which match a provided description such as green tractors or round tables. To solve the CAPTCHA, the user must locate the matching objects. Users are allowed to make one mistake, either selecting an incorrect object or missing one object that is present, and still have their attempt counted as correct.
Results
In testing, humans achieved 94.6% accuracy in solving aiCAPTCHA tests.
Demos
Publications
Conference Papers
- B. M. Powell, E. Kalsy, G. Goswami, M. Vatsa, R. Singh, and A. Noore, “Attack-Resistant aiCAPTCHA using a Negative Selection Artificial Immune System,” in Proceedings of the 38th IEEE Symposium on Security and Privacy, 2nd Workshop on Bio-inspired Security, Trust, Assurance, and Resilience, San Jose, CA, 2017. PDF Publisher's Website