posted onGEEKNEWS2017-12-01 18:08
Earlier this year, Google released the AIY project to allow more people to learn, explore and experience artificial intelligence. The AIY Vioce KIt, released shortly after that, enabled many makers to create more possibilities with voice recognition support.
Today, Google has released the latest achievement of its AIY project, Vision Kit, Google's first project to provide neural network acceleration on the device, providing computer vision support without connecting to the cloud.
Like the AIY Voice Kit released in May this year, the Vision Kit is a DIY kit. The main component of the AIY Vision Kit is the VisionBonnet accessory board for the Raspberry Pi. The component's low-power vision processing unit can run neural network models on-device.
So, what can makers do with AIY Vision Kit? With three TensorFlow-based neural network models provided by Google, DIY devices with Vision Kit can recognize thousands of common objects, recognize facial expressions, or detect photographed people, cats, and dogs. Google also provides the Vision Kit's model compilation tools, so you can train and retrain models with TensorFlow on your workstation or any cloud service.
So, what does it "see": what species is the plant activity photographed? When will your dog show up at the door? Do your guests like the holiday decorations in your home? When did your child break into your room?
Open your mind, let go of your thinking, and let your imagination run wild!
