Google is no stranger to the Raspberry Pi platform, as witnessed by its Artificial Intelligence Yourself (AIY) Projects initiative. Today, the company is launching a new product under the AIY umbrella that is called the Vision Kit.
The AIY Vision Kit includes a VisionBonnet circuit board, RGB button, piezo speaker, macro/wide lens and a cardboard outer shell (among other things). You will have to provide a Raspberry Pi Zero W, a Raspberry Pi Camera, a power supply and an SD card. However, once you have all the necessary components together, the VisionBonnet's onboard Intel Movidius MA2450 vision processing unit is able to get to work.
Using Google's included software, tinkerers will find three neural network models based on Google's TensorFlow open-source software library. These models can recognize ordinary everyday objects, human faces/expressions and even includes a dog/cat detector. If you want to do your own training and retraining, Google provides a tool to compile Vision Kit models.
Google notes that there are a number of tasks that the Vision Kit can accomplish, like letting you know when your dog is standing by the door (waiting for an opportunity go outside), or when a particular person has entered a room (perhaps a room that they shouldn't be in), or identify plants and animals that may be lurking in your yard.
Google's AIY Vision Kit is available for pre-order right now from Micro Center, priced at $44.99. It will be available by December 31st according to the product listing.
Google's first AIY project, the Voice Kit, was released back in May of this year. The Voice Kit provide access to the Google Assistant, making it akin to a budget Google Home AI speaker.