After 8 years of operation, Asirra is shutting down effective October 1, 2014. Thank you to all of our users!
Asirra was a human interactive proof that asks users to identify photos of cats and dogs. It was powered by over three million photos from our unique partnership with Petfinder.com.
Web services are often protected with a challenge that's supposed to be easy for people to solve, but difficult for computers. Such a challenge is often called a CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart) or HIP (Human Interactive Proof). HIPs are used for many purposes, such as to reduce email and blog spam and prevent brute-force attacks on web site passwords.
Today, the most common HIPs ask users to identify text that has been distorted or obscured, like the example seen to the right. Unfortunately, such challenges can be difficult and frustrating for people, yet are often easily solved by computers.
Asirra (Animal Species Image Recognition for Restricting Access) is a HIP that works by asking users to identify photographs of cats and dogs. This task is difficult for computers, but our user studies have shown that people can accomplish it quickly and accurately. Many even think it's fun!
Past projects have used photographs to tell computers and humans apart. Examples include Carnegie Mellon's PIX CAPTCHA, Oli Warner's KittenAuth, and work done by Chew and Tygar. These projects have a common weakness: they use relatively small image databases. There's a fundamental reason for this. It's difficult for a computer to automatically classify pictures with high accuracy — that's why the task is useful as a HIP. An image database small enough to be constructed manually by a researcher is also small enough to be manually reconstructed by an attacker.
Asirra is different because of our unique partnership with Petfinder.com, the world's largest site devoted to finding homes for homeless pets. They've provided us with over three million images of cats and dogs, manually classified by people at thousands of animal shelters across the United States. In exchange, we provide a small "Adopt Me!" link beneath each photo, supporting Petfinder's primary mission of finding homes for homeless animals.
Our work was inspired in part by Frozen Bear's HotCaptcha, which similarly uses a large database of manually classified images as the basis for a HIP. We were motivated to improve on the idea because HotCaptcha isn't really appropriate for widespread deployment. Asking users to rate people's attractiveness is potentially offensive, and the challenge does not have any ground truth.
- Jeremy Elson and Jon Howell, Handling Flash Crowds from your Garage, in Proceedings of the 2008 USENIX Annual Technical Conference, Association for Computing Machinery, Inc., June 2008.
- Jeremy Elson, John R. Douceur, Jon Howell, and Jared Saul, Asirra: A CAPTCHA that Exploits Interest-Aligned Manual Image Categorization, in Proceedings of 14th ACM Conference on Computer and Communications Security (CCS), Association for Computing Machinery, Inc., October 2007.