Google produces some of the best still photography with its Pixel phones, but its method is more than just magic. Instead, it's the result of hard technical labor and lots of machine learning models. When Google introduced a refined portrait mode on the Pixel 3, it utilized a wacky 5-phone case to train its ML models — but the rig the company created to enable its new Portrait Light mode might be even wilder. Read More
Google's research projects have been rolled into google.ai to create the new Google AI division. This means that Google Research is no more, and a new website has been launched alongside a renamed blog. The move unifies all of the company's advanced research efforts while explicitly pointing to the machine learning tech that underpins them. Read More
It's hard enough for us to keep track of who's talking in a loud or crowded party, imagine how difficult it is for automated systems to follow. Speech recognition at a reasonable quality is really only something that's been mastered in the last decade or two, add in conflicting sounds as people talk over each other, and an already tricky problem becomes much harder.
Fortunately (or unfortunately) for us, researchers at Google have been working on isolating sources of audio like speech in videos, and the results they showed off yesterday are kind of incredible and simultaneously terrifying. Read More
When you navigate to a website on your expensive new Android device, or try to view an image that someone has sent you on your gorgeous Super AMOLED Quad HD display, the last thing you want is to find yourself standing there, waiting for a progress bar to crawl across the screen, or to squint angrily at the spinning loading icon as it sputters.
(Did you know that the loading icon is called a “throbber”? I just found out, and I’m now stuck on the idea of a “sputtering throbber.” That’s neither here nor there.) Read More
Get ready for the little person living inside your phone and speaker to sound a lot more life-like. Google believes it has reached a new milestone in the quest to make computer-generated speech indistinguishable from human speech with Tacotron 2, a system that trains neural networks to generate eerily natural-sounding speech from text, and they have the samples to prove it. Read More
Watermarks are the most common way to prevent images from being used without licensing, but they're not nearly as potent as you'd imagine. Research at Google recently published a paper called "On the Effectiveness of Visual Watermarks" regarding how easily watermarks can be removed, and how this can be prevented. Read More
Smartphone cameras have come a long way, but can you ever take images that rival a "real" camera? According to Google software engineer Florian Kainz, the answer is yes. Using a custom camera app and some post-capture editing , Kainz shows what the camera sensors in the Pixel and Nexus 6P can do in low light situations. Read More
A few weeks ago, a new Parking Difficulty icon started showing up in Google Maps 9.44 beta in some cities in the United States, then Google officially announced it and specified where it's available. It also said that the estimate is "based on historical parking data," similar to how traffic and popular times and visit durations are calculated, but it didn't go into a lot of details. Now those details are further clarified in a post on the Google Research Blog.
The difficulties of calculating parking availability stem from the many, many factors that can influence the equation: time and day and weather and holidays/events, permit or illegal parking in park-meter areas, vacant spots with paid park-meters from cars that left early, parking lots with multiple levels and different structures, and so on. Even Read More