{
  "fetched_at": "2026-10-19T19:13:36Z",
  "source_url": "https://www.kaggle.com/competitions/ai2/data",
  "pages": [
    {
      "name": "rules",
      "content": "\n<!-- Begin Base Rules -->\n    <h3>Competition Timeline</h3>\n    <p>Start Date: <strong>3/4/2018</strong></p>\n    <p>End Date: <strong>5/4/2018 23:59 PM Central Time</strong></p>\n    <p>The rules are subject to change. Please check the updates in the forum (Discussion).</p>    \n\n<!-- End Base Rules -->",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "data-description",
      "content": "<p>There are 200 images for training, 100 images for validation and 100 images for test. </p>\n<p>The validation set and the test set are packaged together. For each submission, you need to predict the category of each image in validation set and the test set.\n<p>Your AI is required to predict if there are birds or not at each val/test image in an automatic manner. This is a binary classification task, as predicting the exact number of birds is not required.</p>\n<h2>File descriptions</h2>\n<ul>\n<li><strong>train.zip</strong> - the training set</li>\n<li><strong>val_and_test.zip</strong> - the union of the validation set and the test set</li>\n<li><strong>sample_submission.csv</strong> - a sample submission file in the correct format</li>\n<li><strong>simple_cnn.ipynb</strong> - an example classifier based on convolutional neural network</li>\n<li><strong>extra_larger_dataset.zip</strong> - an larger dataset (released on 5/3/2018) containing 3000 images for optional experiments. The models trained on these images cannot be evaluated by the online evaluation system. You should write your own evaluation program offline. Not only if a image contains a bird but also the exact number of birds are labeled. You can find these information in the csv files.</li>\n</ul>\n<h2>Data fields</h2>\n<ul>\n<li><strong>image</strong> - an unique filename to a given image</li>\n<li><strong>has_bird</strong> - a binary value that indicates if there is at least a bird in the corresponding image</li>\n</ul>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "<h2>Welcome</h2>\n<p>Welcome to Bird Recognition Competition in AI 2 class. In this competition, you are required to recognize the birds from aerial images. Since this competition is considered as final project of AI 2 class, please use the knowledge you studied in the class to finish your system.</p>\n\n<h2>Why recognize bird?</h2>\n<p>The number of birds in each area has a big effect on local life environment around the world. Every year, the birds will migrate from one place to another place in different season and it will affect plant grown and soil moisture. In previous years, people will fly in the sky and use the camera on the air-plane to roughly count how many birds in the image which is not accuracy and time consuming. </p>\n<p>By participating, teams will work to automate the process of identifying birds in the images, which will allow for more efficient way to evaluate the local ecosystem. The final ranking will be based on Category Accuracy of test data.</p>\n\n<h2>Data Description</h2>\n<p>All the images are collected by Missouri Department of Conservation in United States and labeled by <a href=\"http://dslsrv1.rnet.missouri.edu/\">Distributed and Intelligent Computing Lab</a> from University of Missouri. In this competition, all images have been resized into 224*224, you are required to classify the category of each image. There are only two categories in this competition, bird or non-bird. </p>\n\n<h2>What expect?</h2>\n<p>In this competition, you are expected to implement an automatic AI system which have a fairly high accuracy on classifying if there are birds in each image. You can form as a team or finish it as solo competitor. You can use any knowledge you learned from AI 2 class to implement your system. However, in terms of fairness, if you use knowledge uncovered from AI 2 class, please post related information in the discussion board and let other competitor know about it. </p>\n\n\n  [1]: http://dslsrv1.rnet.missouri.edu/",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "<p>This competition will use Category Accuracy as evaluation metrics. The formula of Category Accuracy is as follow:</p>\n<p>\\[ Accuracy = \\frac{TruePositive+ TrueNegative}{Total \\quad number \\quad of \\quad images}\\]</p>\n<p>For more details of True Positive and True Negative, please check wiki about <a href=\"https://en.wikipedia.org/wiki/Confusion_matrix\"> confusion matrix</a></p>\n<h2>Submission Format</h2>\n<p><strong>For every author in the dataset</strong>, submission files should contain two columns: image, has_bird. The file should contain a header and have the following format:</p>\n<pre>image,has_bird<br />1001_G377_15Nov2016_0480.JPG,1<br /><span style=\"font-size: 14px; line-height: 1.4em;\">1002_G377_25OCT2016_0487.JPG,0<br />1003_G377_15Nov2016_0480.JPG,1<br /></span><span style=\"font-size: 14px; line-height: 1.4em;\">1004_G377_19OCT2016_0529.JPG,1<br /></span><span style=\"font-size: 14px; line-height: 1.4em;\">...</span></pre>\n",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}