Google I/O 2017, as one of the most attractive tech events, made many people excited because of the charm of technology. Although there were some boring warm-up games before the keynote, it finally turned out to be worthy of waiting because of high-quality content output during the last two hours. Also, Google I/O took instances to illustrate why it is still the most concerned technical conference now. Here are the major results Google showed on the keynote!
Google Lens: New AI Visual Search Application
One of the centerpieces of the keynote is an AI application, Google Lens, which is developed based on image recognition and OCR. Technically, Google Lens can use image recognition to identity everything appearing in the camera lens in real time. The goal of Google Lens is to make smart phones see the world and even understand the world from camera, so that you can take action more effectively.
Google did not offer a certain date for when this app can be available, but did say that Google Lens will be firstly applied to Google Assistant and Google Photos, and expanded to a wide range of applications in the future.
Google Assistant Finally Will Get on iPhone
Since its launch 2016, Google Assistant now has been used on more than 100 million devices, including Android smart phones, watches, tablet PC, and Google Home. It is a good news for Google, but that is not enough. To make Google Assistant reach more users, Google plans to make it get on iPhone with iOS 9.1 installed.
What’s more, Google released some new features and updates. Literal interactive function now is added on Google Assistance, because Google found there are 70% commands given by natural voice. Google Calendar and Remainder will be available on Google Assistant soon. Users can use Google Assistant to control over more smart home devices. An SDK now is released for Google Assistant, allowing third-party services accessing to Google Assistant. Even better, working with Google Lens above, Google Assistant can help users to recognize the circumjacent information.
Google Home: Big Software Upgrades
Without any hardware upgrade, Google Home brings many new functions. The first one is that the screens of televesions, smartphones and all the devices now can connect to Google Home, instead of the built-in screen of Google Home. Google called it Visual Responses.
When users need screen interaction, they can talk to Google Home and let the result showing on your smartphones. If you want to watch TV, all televisions using Chromecast connectivity will become the device with Google Assistant voice function.
The second large upgrade is Hands-Free Calling, which is long-awaited. This function uses voice recognition to identify that are speaking and making the call. So, different person can use the same voice command to make the call to different persons.
It is obvious that all these upgrades are the challenges to Amazon Echo, and Google Home will be available in more countries including the UK, Canada, Australia, France, Germany, and Japan.
Google Photos: Make Sharing More Intelligent
Google brought 3 new ways of sharing photos to help users share meaningful photos to relatives. One is Suggested Sharing. By utilizing machine learning technology, Google Photos will screen out the right photos and give suggestions to sharing photos. Just clicking on the Send button, the photos will be shared.
The second one is Shared Libraries, which allows users sharing certain photos with specific friends. The last one is Photo Books that Google Photos will pick out the best photos via image recognition and automatically remove the repetitive ones and low-quality photos, and then form a photo album.
All these new functions will be issued in the next couple of weeks.
Google.ai: Display the Best AI
Google announced the second generation of Tensor processing unit (TPU), which is a system for cloud hardware and software. It will get Google Cloud online, namely Cloud TPU, which can boost a large amount of load that machine learning demands, including training and inference. Generally speaking, it will have a good effect on Google cloud platform.
Due to the powerful strength of hardware and software integration, TensorFlow now has become the top platform for building AI software. TPU is the special chip for machine learning. The new generation of TPU can provide up to 180 petaflops.
To make it more powerful, Google designed a custom high-speed network for each TPU, which can be combined to form a machine learning super computer, called TPU POD. Each TPU pod includes 64 second-generation TPU, which turns Google servers into the super computers with 11.5 petaflops. It can accelerate each single large-scale machine learning mode training.
These will be provided for future Google cloud engine, and all AI related content will be displayed on newly issued Google.ai.
Google I/O 2017’s Keynote announced a series of new features and technologies, and most of them are long-awaited, even though there is no the surprising black technologies such as Google glass. The new things above are just the highlights we catch from the keynote. Actually, Google also brought some new and exciting features in terms of VR/AR, Android O, and YouTube. Well, Kotlin becomes the official language of Android is really good to developers. All content showed that Google is stably moving forward.