NeuroDog: Dog Breed Identifier

(C) SnowCron.com

Figuring out the breed of dog by its photo is vital for modern society. Our software answers this formidable challenge, allowing the user to identify 120 (it is one hundred twenty) different breeds and to make educated guess in case of mixed breeds or human photos. It also displays non-intrusive commercials pointing to our other sites: just what the clients need, we are sure of that.

Let's walk through the functional of this wonderful program.

Note: if you are a developer, you will find technical details in this tutorial.

User Interface

The following screenshot shows the initial (right after the program was started) User Interface:

Top - left button is "Tutorial", it will take the user to this page.

Second and third buttons are "File" (Gallery) and "Camera": one can load an image from either camera (if the device has one) or a file.

The leftmost button in a second row is "Commercials": clicking on it will bring up a list of other sites that you should definitely visit. Clicking this button again hides the list.

The last button is... invisible. It is a "Process" button, but to process an image obe should load it first.

Use Cases

There are three ways you can supply an image for analysis.

First, click "Gallery" and select an image containing a dog.

Second, click "Camera", make a photo of a dog and (it depends on your camera software, but mostly is required) click the checkmark, to let the camera program know that you are done making pictures and want to send the last image to NeuroDog.

Third, you can "share" the image from some applications, like Facebook. To do it, look for the "share" icon:
Note: NeuroDog will be shown in the list of applications that can handle the shared image, but not for all applications. For example, Web Browser usually shares an HTML page, which NeuroDog can not handle.

After you have chosen an image, it will be displayed in the app. and the "Process" button will be displayed:

Press the "Process" button to analyze the image:

The Neural Network our application is based on will think for some time, and display five most probable candidate breeds, together with their probabilities. Maximum probability is 1, usually anything below 0.9 is a mixed breed (or a poor quality photo):

Good luck :)