This is the TechSummit Rewind, a daily recap of the top technology headlines.
Google using machine learning to reduce data needed for high-resolution images
Google has begun the process of integrating RAISR, a prototype technology that uses machine learning to make low-resolution images appear more detailed, into its online services. The technology will first upscale large images on Google+ and save users’ data in the process.
So far, it’s being used to tweak high-resolution Google+ images accessed on a “subset of Android devices.” When a user requests an image, Google+ retrieves a version that’s actually a quarter of the size and uses RAISR’s algorithms to “restore detail on [the] device.” This reduces the data cost of each image by up to 75 percent, according to Google. This technique is currently being applied to over a billion images a week, and has reduced “users’ total bandwidth by about a third,” according to Google.
The technique inserts new pixels into low-resolution images to make up for lost detail. However, while traditional upscaling uses fixed rules to work out which new pixels to use where, RAISR adapts its method to each image. It pays special attention to “edge features” (bits of the image that look like the edge of an object), making the resulting larger image look less blurred.
Intel wants sensors to help you with your shopping
Intel is launching a Responsible Retail Platform that creates a common set of sensors, software kits, and other components for in-store tech. It promises to speed up inventory tracking, provide feedback on buying habits and personalize your shopping.
The company is investing over $100 million into the retail industry over the next five years to help get the ball rolling.
Raspberry Pi Compute Module 3 launches at $30, ready to power up in other products
Raspberry Pi Foundation has launched the Compute Module 3, a slimmed-down Raspberry Pi 3 for developing customized hardware like TV displays, industrial control systems, and home media players.
It delivers double the RAM and a 10-times boost to CPU performance over the first iteration released in 2014.
The new module contains the Raspberry Pi 3’s guts, featuring a 64-bit Broadcom BCM2837 processor at up to 1.2GHz with 1GB of RAM and 4GB of flash storage. By removing a number of ports, it offers a smaller design to make it more suitable than the Raspberry Pi for industrial applications.
The device also fits the DDR2 SODIMM sockets and has the same pin-out as the first Pi compute module. One notable change is that the Compute Module 3, at 31mm high, is one mm taller than the original one.
“The idea of the Compute Module was to provide an easy and cost-effective route to producing customized products based on the Pi hardware and software platform.
“The thought was to provide the ‘team in a garage’ with easy access to the same technology as the big guys. The Module takes care of the complexity of routing out the processor pins, the high-speed RAM interface, and core power supply, and allows a simple carrier board to provide just what is needed in terms of external interfaces and form factor.”
-James Adams, Raspberry Pi Foundation COO
The Foundation has also launched the Compute Module 3 Lite, which has the same processor and 1GB of RAM minus the flash storage. Builders can add an eMMC device or SD card socket on the base board.
Lastly, it’s releasing an updated version of the Compute Module IO Board 3, which provides HDMI and USB connectors, powers the module, and enables programming of its flash memory.
The standard Compute Module 3 costs $30 while the Lite version will set you back $25, excluding shipping and tax. The original Compute Module has been reduced to $25 from distributors RS Components, element 14, and Farnell UK.
Showing off its commercial applications, the Raspberry Pi Foundation highlighted a partnership with the European arm of NEC to integrate the module into a line of new large-format displays.
The module will power signage software and presentations on the NEC displays, and for CubeSats (mini-satellites made with off-the-shelf technology) by UK researchers.
Xiaomi stops disclosing annual sales figures, CEO admits company grew too fast
Xiaomi has forgone its tradition of revealing how many smartphones the company sold the previous year. Its CEO Lei Jung admitted Xiaomi has been in transition after growing “too fast.”
“In the first few years, we pushed ahead too fast. We created a miracle, but also drew on some long-term growth. So we have to slow down, further improve in some areas, and ensure sustainable growth for a long-term future.”
-Lei Jung; Xiaomi CEO, in a letter to employees
However, Lei told his staff that the “difficult times are behind us.”
A range of business metrics were also provided in Lei’s letter that give a picture of how the company is doing:
- Xiaomi reached $1 billion in annual revenue in India for the first time
- Three of Xiaomi’s 54 offline stores passed $14.5M in gross merchandise value each – including sales of partner products – and it plans to open 200 more in 2017, and 1,000 more by 2020
- Xiaomi’s ‘Mi Ecosystem’ has 50 million connected devices – which includes smart TVs, fitness trackers, and more – which brought in $2.2 billion in sales
- Xiaomi has applied for 16,000 patents globally, 3,612 of which have been granted to date
- Xiaomi’s revenue from “internet services” doubled in 2016
The most interesting note here is that Lei wants to push on and see Xiaomi develop its offline retail arm.
According to Lei, Xiaomi has to branch out because its current distribution model is limited.
“Xiaomi has great ambitions, and we are not satisfied with just being an e-commerce smartphone brand, so we have to upgrade our retail model, and incorporate offline retail for a new retail strategy.”
Xiaomi also plans on developing artificial intelligence, increasing its global presence, and developing fintech solutions in 2017.
Jung’s “humble” goal for the year is to reach $14.5 billion in revenue.
Facebook rolls out fake news filter in Germany
Facebook will begin rolling out its fake news filter in Germany, according to the Financial Times, amid lawmakers expressing growing concern over the spread of fabricated news stories and Russian interference ahead of its national elections later this year.
The social network will begin fact-checking and flagging fake news for users in Germany over the coming weeks.
Under Facebook’s fact-checking system, stories reported as fake by users will be sent to Correctiv, a nonprofit news organization based in Berlin. If an item is deemed false, it will be marked as “disputed,” along with justification, and the site will warn users before they share it. Disputed items will also show up lower in News Feeds.
“Our focus is on Germany right now but we’re certainly thinking through what countries will unveil next.”
-A Facebook spokesperson
The Bragi Headphone starts shipping
Bragi’s latest Headphone earbuds are now available for immediate shipping after a series of delays.
The earbuds are a stripped down version of the company’s Dash earbud, and ditches the fitness tracking, heart rate monitoring, gesture controls, and app to simply function as a pair of truly wireless Bluetooth headphones. It’s shipping now on Bragi’s website now for $149.99, with Bragi estimating orders to arrive within two weeks.