Google today announced a partnership with Silicon Valley chip maker Movidius to help bring more powerful image recognition technology directly to smartphones. In current devices, those kinds of capabilities are typically restricted to cloud-based apps, which communicate with servers to tap into advanced artificial intelligence techniques for performing facial recognition and other tasks. The new collaboration, however, will see Google place Movidius' MA2450 chip inside Android handsets to allow them to understand and identify images like faces and street signs in real time, without the need to upload photos and wait for algorithms to do their magic in the cloud.
The potential is huge for people with disabilities — an image-recognizing smartphone could help blind individuals more easily maneuver urban areas or converse with familiar people on the street. It's an extension of what Google already does with Google Translate, which works entirely offline and allows users to hold their phone in front of words in a foreign language and see the text translated on the screen. This time, though, Google is trying to crack the difficult problem of recognizing objects and even people, not just text. Google Photos, the company's storage software, already does this in the cloud, letting you search for things like mountains or flowers online and see corresponding photos, but only when connected to Google servers.
The next generation of smartphones will be able to see on their own
Movidius, which specializes in computer vision, has worked with Google in the past on Project Tango, the company's initiative aimed at allowing phones to 3D map real-world environments. "By working with Movidius, we’re able to expand this technology beyond the data center and out into the real world, giving people the benefits of machine intelligence on their personal devices," Blaise Agϋera y Arcas, Google's head of machine intelligence, said in a video. Movidius chief Remi El-Ouazzane says Google will launch a new generation of devices with Movidius' technology in the "not-too-distant-future."