{
  "fetched_at": "2026-10-19T07:46:40Z",
  "source_url": "https://www.kaggle.com/competitions/dogs-vs-cats/data",
  "pages": [
    {
      "name": "rules",
      "content": "\r\n\t<!-- Begin Base Rules -->\r\n    <h3>One account per participant</h3>\r\n    <p>You cannot sign up to Kaggle from multiple accounts and therefore you cannot submit from multiple accounts.</p>\r\n    <h3>No private sharing outside teams</h3>\r\n    <p>\r\n        Privately sharing code or data outside of teams is not permitted.\r\n        It's okay to share code if made available to all participants on the forums.\r\n    </p>\r\n    <h3>Team Mergers</h3>\r\n    <p>\r\n        Team mergers are allowed and can be performed by the team leader. In order to merge, the combined team must have a total submission count less than or equal to the maximum allowed as of the merge date. The maximum allowed is the number of submissions per day multiplied by the number of days the competition has been running.<br/>\r\n    </p>\r\n\r\n    <h3>Team Limits</h3>\r\n    <p>There is no maximum team size.</p>\r\n    <h3>Submission Limits</h3>\r\n    <p>You may submit a maximum of 2 entries per day.</p>\r\n    <p>You may select up to 2 final submissions for judging.</p>\r\n    <h3>Competition Timeline</h3>\r\n    <p>Start Date: <strong>9/25/2013 2:44 PM UTC</strong></p>\r\n    <p>Merger Deadline: <strong>None</strong></p>\r\n    <p>Entry Deadline: <strong>None</strong></p>\r\n    <p>End Date: <strong>2/1/2014 11:59 PM UTC</strong></p>\r\n\t<!-- End Base Rules -->\r\n<ul>\r\n<li><strong><span style=\"font-size: 1em; line-height: 1.5em;\">HAND LABELING IMAGES IS STRICTLY FORBIDDEN.</span></strong></li>\r\n<li><span style=\"font-size: 1em; line-height: 1.5em;\">External data is allowed, but must be posted in the forums for all to see and use.</span></li>\r\n<li><span style=\"color: #555555; font-size: 1em; line-height: 1.5em;\">These images have been published by Microsoft Research for the express purpose of furthering academic research. They may be used for non-commercial research purposes, but they may not be re-published without the express permission of Microsoft Research.</span></li>\r\n</ul>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "<p>In this competition, you'll write an algorithm to classify whether images contain either a dog or a cat. &nbsp;This is easy for humans, dogs, and cats. Your computer will find it a bit more difficult.</p>\r\n<p style=\"text-align: center;\"><img src=\"https://storage.googleapis.com/kaggle-media/competitions/kaggle/3362/media/woof_meow.jpg\" alt=\"Woof Meow\" width=\"450\" height=\"278\" /></p>\r\n<p style=\"text-align: center;\">Deep Blue beat Kasparov at chess in 1997.<br />Watson beat the brightest trivia minds at Jeopardy in 2011.<br />Can you tell Fido from Mittens in 2013?</p>\r\n<h2>The Asirra data set</h2>\r\n<p>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&nbsp;<a href=\"http://www.captcha.net/\" target=\"_new\">CAPTCHA</a>&nbsp;(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.</p>\r\n<p><a href=\"http://research.microsoft.com/en-us/um/redmond/projects/asirra/\">Asirra</a> (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 studies&nbsp;have shown that people can accomplish it quickly and accurately. Many even think it's fun! Here is an example of the Asirra interface:</p>\r\n<form id=\"mainForm\" action=\"\" method=\"get\" onsubmit=\"\">\r\n<script type=\"mce-text/javascript\" src=\"https://challenge.asirra.com/js/AsirraClientSide.js\"></script>\r\n<script type=\"mce-text/javascript\">// <![CDATA[\r\n// You can control where the big version of the photos appear by\r\n\t\t\t// changing this to top, bottom, left, or right\r\n\t\t\tasirraState.SetEnlargedPosition(\"right\");\r\n\r\n\t\t\t// You can control the aspect ratio of the box by changing this constant\r\n\t\t\tasirraState.SetCellsPerRow(6);\r\n// ]]></script>\r\n</form>\r\n<p>Asirra is unique because of its partnership with&nbsp;<a href=\"http://www.petfinder.com/\" target=\"_new\">Petfinder.com</a>, the world's largest site devoted to finding homes for homeless pets. They've provided Microsoft Research with over&nbsp;three million images&nbsp;of cats and dogs, manually classified by people at thousands of animal shelters across the United States. Kaggle is fortunate to offer a subset of this data for fun and research.&nbsp;</p>\r\n<h2>Image recognition attacks</h2>\r\n<p>While random guessing is the easiest form of attack, various forms of image recognition can allow an attacker to make guesses that are better than random. There is enormous diversity in the photo database (a wide variety of backgrounds, angles, poses, lighting, etc.), making accurate automatic classification difficult. In an informal poll conducted many years ago, computer vision experts posited that a classifier with better than 60% accuracy would be difficult without a major advance in the state of the art. For reference, a 60% classifier improves the guessing probability of a 12-image HIP from 1/4096 to 1/459.</p>\r\n<h2>State of the art</h2>\r\n<p>The current literature suggests machine classifiers can score above 80% accuracy on this task <a href=\"http://xenon.stanford.edu/~pgolle/papers/dogcat.pdf\">[1]</a>. Therfore, Asirra is no longer considered safe from attack. &nbsp;We have created this contest to benchmark the latest computer vision and deep learning approaches to this problem. Can you crack the CAPTCHA? Can you improve the state of the art? Can you create lasting peace between cats and dogs?</p>\r\n<p>Okay, we'll settle for the former.&nbsp;</p>\r\n<h2>Acknowledgements</h2>\r\n<p>We extend our thanks to Microsoft Research for providing the data for this competition.</p>\r\n<ul>\r\n<li><span style=\"font-size: 1em; line-height: 1.5em;\">Jeremy Elson, John R. Douceur, Jon Howell, 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., Oct. 2007</span></li>\r\n</ul>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Prizes",
      "content": "<ul>\r\n<li>1st place - $126 donation to the ASPCA or an animal charity of your choosing</li>\r\n<li>2nd place - A stuffed <a href=\"http://amzn.com/B006U4SHMS\">Boo</a></li>\r\n</ul>\r\n<p><img src=\"https://storage.googleapis.com/kaggle-media/competitions/kaggle/3362/media/boo.jpg\" alt=\"\" width=\"200\" height=\"200\" /></p>\r\n<!--<h2>Donations</h2>\r\n<p><iframe id=\"JotFormIFrame\" onload=\"window.parent.scrollTo(0,0)\" allowtransparency=\"true\" src=\"https://secure.jotformpro.com/form/32675791716970\" frameborder=\"0\" style=\"width: 100%; height: 465px; border: none;\" scrolling=\"no\"></iframe></p>\r\n<script type=\"mce-text/javascript\">// <![CDATA[\r\nwindow.handleIFrameMessage = function(e) {var args = e.data.split(\":\");var iframe = document.getElementById(\"JotFormIFrame\");if (!iframe)return;switch (args[0]) {case \"scrollIntoView\":iframe.scrollIntoView();break;case \"setHeight\":iframe.style.height = args[1] + \"px\";break;}};if (window.addEventListener) {window.addEventListener(\"message\", handleIFrameMessage, false);} else if (window.attachEvent) {window.attachEvent(\"onmessage\", handleIFrameMessage);}\r\n// ]]></script>\r\n-->",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "<p>Performance is evaluated on the percentage of correctly labeled images. To determine your odds of breaking the Asirra CAPTCHA, raise your percentage to the 12th power.</p>\r\n<p>\"But classification accuracy is a flawed metric!\" you scream at your monitor in fury, \"my genius requires you accept the posterior probability of my predictions!\" &nbsp;That may be true, but sometimes simplicity is just nice. Here there are only dogs and cats... no 0.5 dog-cat hybrid guesses allowed!</p>\r\n<h2>Submission Format</h2>\r\n<p>Your submission should have a header. For each image in the test set, predict a label for its id (1 = dog, 0 = cat):</p>\r\n<pre>id,label\r\n1,0\r\n2,0\r\n3,0\r\netc...\r\n</pre>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "data-description",
      "content": "<p>The training archive contains 25,000 images of dogs and cats. Train your algorithm on these files and predict the labels for test1.zip (1 = dog, 0 = cat).</p>\n<h2>A note on hand labeling</h2>\n<p>Per the rules and spirit of this contest, please do not manually label your submissions. We work hard to fair and fun contests, and ask for the same respect in return.</p>\n<p style=\"text-align: center;\">&nbsp;<img title=\"Please don't cheat\" src=\"https://storage.googleapis.com/kaggle-media/competitions/kaggle/3362/media/please_dont_cheat.jpg\" alt=\"Please don't cheat\" width=\"300\" height=\"226\" /><br /> \"I'm a good doggie who deserves a good algorithm.<br />And chew bones. <em>Please send chew bones</em>.\"</p>\n<p>However, given that there is always one willing to cheat to rise to glory, <strong>we reserve the right to wield the following anti-doping measures</strong>:</p>\n<ul>\n<li>Ask to spot check your code at any point in the competition. Failure to provide proof of algorithm within a reasonable time frame may result in disqualification.</li>\n<li>Release a new test set towards the end of the competition and require predictions for this new set&nbsp;<em>(Asirra has 3 million images, lest you think we'll run out of cats).</em></li>\n<li>At any time during the competition, release a randomized version of a test set. This may include nefarious tricks to make matching the old and new test sets very annoying (e.g. removing rows, reversing orientation, rescaling the images, nonlinear pixel transformations).</li>\n<li>After the competition, require access to your model to verify that the solution does not utilize human intervention.</li>\n</ul>\n<div><strong>In short, if you cheat by hand labeling your predictions, expect to do it all over again, possibly many times.</strong> &nbsp;Thanks in advance for a fair competition and have fun.</div>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Winners",
      "content": "<p><strong>First Place:</strong></p>\r\n<ul>\r\n<li>Pierre Sermanet - New York, USA</li>\r\n</ul>\r\n<p><strong>Second Place:</strong></p>\r\n<ul>\r\n<li>Zhao Xing - Livingston, USA</li>\r\n</ul>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}