Cat Gender Classification

Description
<b id="docs-internal-guid-be0bdc16-2c9e-5ce3-148f-0d377ebd4ac1"></b><p dir="ltr"><span style="color: #000000;background-color: transparent;">Given a photo of a cat, the algorithm must determine the gender of the feline.</span></p><br/><p dir="ltr"><span style="color: #000000;background-color: transparent;">Research on </span><a href="https://en.wikipedia.org/wiki/Cat_coat_genetics"><span style="color: #1155cc;background-color: transparent;">cat coat genetics</span></a><span style="color: #000000;background-color: transparent;"> have reached a mature stage, allowing observers to conclude attributes by analyzing the color, pattern, length, and texture. More specifically, cats have been determined to have </span><a href="http://biology.clc.uc.edu/Courses/bio105/sex-link.htm"><span style="color: #1155cc;background-color: transparent;">sex-linked genes</span></a><span style="color: #000000;background-color: transparent;"> that play a direct role with the coloring of their fur. In general, a cat’s fur coat color and length can be used to determine the cat’s breed, and based on the pattern of the colors and the type of breed one could determine the gender with reasonable confidence.</span></p><br/><p dir="ltr"><span style="color: #000000;background-color: transparent;">Note however that the described flow above (coat color -&gt; breed -&gt; pattern -&gt; gender) is not applicable to all breeds of cat. The algorithm developer must accommodate the outliers whenever it is possible to automate.</span></p><br/><p dir="ltr"><span style="color: #000000;background-color: transparent;"><b>Input</b></span></p><p dir="ltr"><span style="color: #000000;background-color: transparent;">The input must accept 1) DataAPI path to an image, and 2) URL to an image. It would be preferable to allow additional input options, such as Base64 encoding of the image or binary.</span></p><br/><p dir="ltr"><span style="color: #000000;background-color: transparent;"><b>Output</b></span></p><p dir="ltr"><span style="color: #000000;background-color: transparent;">The output must be in the form of a JSON object with at least two elements: gender (‘male’, ‘female’) and confidence (0 to 1). Given how dependent this method is with the genetic profiling of the cat, it would be preferable to include additional information, such as breed, color, and pattern..</span></p><br/><p dir="ltr"><span style="color: #000000;background-color: transparent;"><b>Accuracy and Assumptions</b></span></p><p dir="ltr"><span style="color: #000000;background-color: transparent;">It is fair for the algorithm writer to assume that all input images provide ‘adequate exposure’ of the feline, meaning that the image will be taken from the side or top of the cat as opposed to a front/back photo.</span></p><br/><p dir="ltr"><span style="color: #000000;background-color: transparent;">The algorithm must pass 80% of any randomly selected images that meet the adequate exposure criteria.</span></p><br/><p dir="ltr"><span style="color: #000000;background-color: transparent;"><b>Examples</b></span></p><p dir="ltr"><img src="http://i.imgur.com/bKOl7GB.jpg"/><br/></p><p dir="ltr"><span style="color: #000000;background-color: transparent;"><i>Output</i>: {result: ‘female’, confidence: 1, breed: ‘Torbie’}</span></p><br/><p dir="ltr"><img alt="" src="http://i.imgur.com/EI83bi2.jpg"/><br/></p><p dir="ltr"><span style="color: #000000;background-color: transparent;"><i>Output</i>: {result: ‘male’, confidence: 1, breed: ‘Burmilla’, mood: ‘angry’}</span></p><br/><p dir="ltr"><img alt="" src="http://i.imgur.com/YL2EGK9.jpg"/><br/></p><p dir="ltr"><span style="color: #000000;background-color: transparent;"><i>Output</i>: {result: ‘unknown’, confidence: 0, error: ‘inadequate exposure’}</span></p>
Discussion
  • {{comment.username}}
Status
Active
submission(s) pending review
Bounty expires in
Bounty expired
Bounty
100
Tags
(no tags)