Microsoft Cognitive Services

I recently discovered the Microsoft Cognitive Services API, which automates some common AI operations by exposing appropriate web services. These are free to try, but would require payment for usage of any significant scale. The website also allows users to demo a few of these APIs:

Here is me enjoying a noodle dish (though the Emotion API says I was only neutral about it):


The API did better with general feature detection on the image, it classified it as ‘a man sitting at a table with a plate of food’:assaf_eating_features

There are many possibilities for utilizing these API’s, a couple of my favorite examples from the Microsoft site are:

  • Fetch – identify a dog’s breed using a photo
  • Video Breakdown – automatically index (and build a transcript) for a video


Next Steps:

  • Use the API with a Raspberry Pi application (maybe a device that alerts if a room is messy?)
  • Identify open source alternatives to the Microsoft API