Probability Comparisons
31 October 2020
TikTok stars got a judge to block Trump’s TikTok ban
TikTok has won another battle in its fight against the Trump administration’s ban of its video-sharing app in the U.S. — or, more accurately in this case, the TikTok community won a battle. On Friday, a federal judge in Pennsylvania issued an injunction that blocked the restrictions that would have otherwise blocked TikTok from operating in the U.S. on November 12.
This particular lawsuit was not led by TikTok itself, but rather a group of TikTok creators who use the app to engage with their million-plus followers.
According to the court documents, plaintiff Douglas Marland has 2.7 million followers on the app; Alec Chambers has 1.8 million followers; and Cosette Rinab has 2.3 million followers. The creators argued — successfully as it turns out — that they would lose access to their followers in the event of a ban, as well as the “professional opportunities afforded by TikTok.” In other words, they’d lose their brand sponsorships — meaning, their income.
This is not the first time that the U.S. courts have sided with TikTok to block the Trump administration’s proposed ban over the Chinese-owned video sharing app. Last month, a D.C. judge blocked the ban that would have removed the app from being listed in U.S. app stores run by Apple and Google.
That ruling had not, however, stopped the November 12 ban that would have blocked companies from providing internet hosting services that would have allowed TikTok to continue to operate in the U.S.
The Trump administration had moved to block the TikTok app from operating in the U.S. due to its Chinese parent company, ByteDance, claiming it was a national security threat. The core argument from the judge in this ruling was the “Government’s own descriptions of the national security threat posed by the TikTok app are phrased in the hypothetical.”
That hypothetical risk was unable to be stated by the government, the judge argued, to be such a risk that it outweighed the public interest. The interest, in this case, was the more than 100 million users of TikTok and the creators like Marland, Chambers and Rinab that utilized it to spread “informational materials,” which allowed the judge to rule that the ban would shut down a platform for expressive activity.
“We are deeply moved by the outpouring of support from our creators, who have worked to protect their rights to expression, their careers, and to help small businesses, particularly during the pandemic,” said Vanessa Pappas, Interim Global Head of TikTok, in a statement. “We stand behind our community as they share their voices, and we are committed to continuing to provide a home for them to do so,” she added.
The TikTok community coming to the rescue on this one aspect of the overall TikTok picture just elevates this whole story. Though the company has been relatively quiet through this whole process, Pappas has thanked the community several times for its outpouring of support. Though there were some initial waves of “grief” on the app with creators frantically recommending people follow them on other platforms, that has morphed over time into more of a “let’s band together” vibe. This activity coalesced around a big swell in voting advocacy on the platform, where many creators are too young to actually participate but view voting messaging as their way to participate.
TikTok has remained active in the product department through the whole mess, shipping elections guides and trying to ban QAnon conspiracy spread, even as Pakistan banned and then un-banned the app.
Read Full Article
TikTok stars got a judge to block Trump’s TikTok ban
TikTok has won another battle in its fight against the Trump administration’s ban of its video-sharing app in the U.S. — or, more accurately in this case, the TikTok community won a battle. On Friday, a federal judge in Pennsylvania has issued an injunction that blocked the restrictions that would have otherwise blocked TikTok from operating in the U.S. on November 12.
This particular lawsuit was not led by TikTok itself, but rather a group of TikTok creators who use the app to engage with their million-plus followers.
According to the court documents, plaintiff Douglas Marland has 2.7 million followers on the app; Alec Chambers has 1.8 million followers; and Cosette Rinab has 2.3 million followers. The creators argued – successfully as it turns out — that they would lose access to their followers in the event of a ban, as well as the “professional opportunities afforded by TikTok.” In other words, they’d lose their brand sponsorships — meaning, their income.
This is not the first time that the U.S. courts have sided with TikTok to block the Trump administration’s proposed ban over the Chinese-owned video sharing app. Last month, a D.C. judge blocked the ban that would have removed the app from being listed in U.S. app stores run by Apple and Google.
That ruling had not, however, stopped the Nov. 12 ban that would have blocked companies from providing internet hosting services that would have allowed TikTok to continue to operate in the U.S.
The Trump administration had moved to block the TikTok app from operating in the U.S. due to its Chinese parent company, ByteDance, claiming it was a national security threat. The core argument from the judge in this ruling was the “Government’s own descriptions of the national security threat posed by the TikTok app are phrased in the hypothetical.”
That hypothetical risk was unable to be stated by the Government, the Judge argued, to be such a risk that it outweighed the public interest. The interest, in this case, was the over 100 million users of TikTok and the creators like Marland, Chambers and Rinab that utilized it to spread “informational materials,” which allowed the Judge to rule that the ban would shut down a platform for expressive activity.
“We are deeply moved by the outpouring of support from our creators, who have worked to protect their rights to expression, their careers, and to help small businesses, particularly during the pandemic,” said Vanessa Pappas, Interim Global Head of TikTok, in a statement. “We stand behind our community as they share their voices, and we are committed to continuing to provide a home for them to do so,” she added.
The TikTok community coming to the rescue on this one aspect of the overall TikTok picture just elevates this whole story. Though the company has been relatively quiet through this whole process, Pappas has thanked the community several times for its outpouring of support. Though there were some initial waves of ‘grief’ on the app with creators frantically recommending people follow them on other platforms, that has morphed over time into more of a ‘let’s band together’ vibe. This activity coalesced around a big swell in voting advocacy on the platform, where many creators are too young to actually participate but view voting messaging as their way to participate.
TikTok has remained active in the product department through the whole mess, shipping elections guides and trying to ban Qanon conspiracy spread. Even as Pakistan banned and then un-banned the app.
Read Full Article
Facebook hits pause on algorithmic recommendations for political and social issue groups
With just days to go before the U.S. election, Facebook quietly suspended one of its most worrisome features.
During Wednesday’s Senate hearing, Senator Ed Markey asked Facebook CEO Mark Zuckerberg about reports that his company has long known its group recommendations push people toward more extreme content. Zuckerberg responded that the company had actually disabled that feature for certain groups — a fact Facebook had not previously announced.
“Senator, we have taken the step of stopping recommendations in groups for all political content or social issue groups as a precaution for this,” Zuckerberg told Markey.
TechCrunch reached out to Facebook with questions about what kind of groups would be affected and how long the recommendations would be suspended at the time but did not receive an immediate response. Facebook first confirmed the change to BuzzFeed News on Friday.
“This is a measure we put in place in the lead up to Election Day,” Facebook spokesperson Liz Bourgeois told TechCrunch in an email. “We will assess when to lift them afterwards, but they are temporary.”
The cautionary step will disable recommendations for political and social issue groups as well as any new groups that are created during the window of time. Facebook declined to provide additional details about the kinds of groups that will and won’t be affected by the change or what went into the decision.
Researchers who focus on extremism have long been concerned that algorithmic recommendations on social networks push people toward more extreme content. Facebook has been aware of this phenomenon since at least 2016, when an internal presentation on extremism in Germany observed that “64% of all extremist group joins are due to our recommendation tools.” In light of the feature’s track record, some anti-hate groups celebrated Facebook’s decision to hit the pause button Friday.
“It’s good news that Facebook is disabling group recommendations for all political content or social issue groups as a precaution during this election season. I believe it could result in a safer experience for users in this critical time,” Anti-Defamation League CEO Jonathan A. Greenblatt told TechCrunch. “And yet, beyond the next week, much more needs to be done in the long term to ensure that users are not being exposed to extremist ideologies on Facebook’s platforms.”
On Facebook, algorithmic recommendations can usher users flirting with extreme views and violent ideas into social groups where their dangerous ideologies can be amplified and organized. Before being banned by the social network, the violent far-right group the Proud Boys relied on Facebook groups for its relatively sophisticated national recruitment operation. Members of the group that plotted to kidnap Michigan Governor Gretchen Whitmer also used Facebook groups to organize, according to an FBI affidavit.
While it sounds like Facebook’s decision to toggle off some group recommendations is temporary, the company has made an unprecedented flurry of choices to limit dangerous content in recent months, possibly in fear that the 2020 election will again plunge it into political controversy. Over the last three months alone, Facebook has cracked down on QAnon, militias and language used by the Trump campaign that could result in voter intimidation — all surprising postures considering its longstanding inaction and deep fear of decisions that could be perceived as partisan.
After years of relative inaction, the company now appears to be taking seriously some of the extremism it has long incubated, though the coming days are likely to put its new set of protective policies to the test.
Read Full Article
iPhones can now tell blind users where and how far away people are
Apple has packed an interesting new accessibility feature into the latest beta of iOS: a system that detects the presence of and distance to people in the view of the iPhone’s camera, so blind users can social distance effectively, among many other things.
The feature emerged from Apple’s ARKit, for which the company developed “people occlusion,” which detects people’s shapes and lets virtual items pass in front of and behind them. The accessibility team realized that this, combined with the accurate distance measurements provided by the lidar units on the iPhone 12 Pro and Pro Max, could be an extremely useful tool for anyone with a visual impairment.
Of course during the pandemic one immediately thinks of the idea of keeping six feet away from other people. But knowing where others are and how far away is a basic visual task that we use all the time to plan where we walk, which line we get in at the store, whether to cross the street, and so on.
The new feature, which will be part of the Magnifier app, uses the lidar and wide-angle camera of the Pro and Pro Max, giving feedback to the user in a variety of ways.
The lidar in the iPhone 12 Pro shows up in this infrared video. Each dot reports back the precise distance of what it reflects off of.
First, it tells the user whether there are people in view at all. If someone is there, it will then say how far away the closest person is in feet or meters, updating regularly as they approach or move further away. The sound corresponds in stereo to the direction the person is in the camera’s view.
Second, it allows the user to set tones corresponding to certain distances. For example, if they set the distance at six feet, they’ll hear one tone if a person is more than six feet away, another if they’re inside that range. After all, not everyone wants a constant feed of exact distances if all they care about is staying two paces away.
The third feature, perhaps extra useful for folks who have both visual and hearing impairments, is a haptic pulse that goes faster as a person gets closer.
Last is a visual feature for people who need a little help discerning the world around them, an arrow that points to the detected person on the screen. Blindness is a spectrum, after all, and any number of vision problems could make a person want a bit of help in that regard.
The system requires a decent image on the wide-angle camera, so it won’t work in pitch darkness. And while the restriction of the feature to the high end of the iPhone line reduces the reach somewhat, the constantly increasing utility of such a device as a sort of vision prosthetic likely makes the investment in the hardware more palatable to people who need it.
This is far from the first tool like this — many phones and dedicated devices have features for finding objects and people, but it’s not often that it comes baked in as a standard feature.
People detection should be available to iPhone 12 Pro and Pro Max running the iOS 14.2 release candidate that was just made available today. Details will presumably appear soon on Apple’s dedicated iPhone accessibility site.
Read Full Article
iPhones can now tell blind users where and how far away people are
Apple has packed an interesting new accessibility feature into the latest beta of iOS: a system that detects the presence of and distance to people in the view of the iPhone’s camera, so blind users can social distance effectively, among many other things.
The feature emerged from Apple’s ARKit, for which the company developed “people occlusion,” which detects people’s shapes and lets virtual items pass in front of and behind them. The accessibility team realized that this, combined with the accurate distance measurements provided by the lidar units on the iPhone 12 Pro and Pro Max, could be an extremely useful tool for anyone with a visual impairment.
Of course during the pandemic one immediately thinks of the idea of keeping six feet away from other people. But knowing where others are and how far away is a basic visual task that we use all the time to plan where we walk, which line we get in at the store, whether to cross the street, and so on.
The new feature, which will be part of the Magnifier app, uses the lidar and wide-angle camera of the Pro and Pro Max, giving feedback to the user in a variety of ways.
The lidar in the iPhone 12 Pro shows up in this infrared video. Each dot reports back the precise distance of what it reflects off of.
First, it tells the user whether there are people in view at all. If someone is there, it will then say how far away the closest person is in feet or meters, updating regularly as they approach or move further away. The sound corresponds in stereo to the direction the person is in the camera’s view.
Second, it allows the user to set tones corresponding to certain distances. For example, if they set the distance at six feet, they’ll hear one tone if a person is more than six feet away, another if they’re inside that range. After all, not everyone wants a constant feed of exact distances if all they care about is staying two paces away.
The third feature, perhaps extra useful for folks who have both visual and hearing impairments, is a haptic pulse that goes faster as a person gets closer.
Last is a visual feature for people who need a little help discerning the world around them, an arrow that points to the detected person on the screen. Blindness is a spectrum, after all, and any number of vision problems could make a person want a bit of help in that regard.
The system requires a decent image on the wide-angle camera, so it won’t work in pitch darkness. And while the restriction of the feature to the high end of the iPhone line reduces the reach somewhat, the constantly increasing utility of such a device as a sort of vision prosthetic likely makes the investment in the hardware more palatable to people who need it.
This is far from the first tool like this — many phones and dedicated devices have features for finding objects and people, but it’s not often that it comes baked in as a standard feature.
People detection should be available to iPhone 12 Pro and Pro Max running the iOS 14.2 release candidate that was just made available today. Details will presumably appear soon on Apple’s dedicated iPhone accessibility site.
Read Full Article
Q3 earnings find Apple and Google looking to the future for hardware rebounds
“5G is a once-in-a-decade kind of opportunity,” Tim Cook told the media during the Q&A portion of Apple’s Q3 earnings call. “And we could not be more excited to hit the market exactly when we did.”
The truth of the matter is its timing was a mixed bag. Apple was, by some accounts, late to 5G. By the time the company finally announced that it was adding the technology across its lineup of iPhone 12 variants, much of its competition had already beat the company to the punch. Of course, that’s not a huge surprise. Apple’s strategy is rarely a rush to be first.
5G networks are only really starting to come into their own now. Even today, there are still wide swaths of users who will have to default to an LTE connection the majority of the time they use their handsets. The arrival of 5G on the iPhone was really as much about future-proofing this year’s models as anything. Consumers are holding onto phones longer, and in the three or four years before it’s time for another upgrade, the 5G maps will look very different.
Clearly, the new iPhone didn’t hit the market exactly when Apple had hoped; the pandemic saw to that. Manufacturing bottlenecks in Asia delayed the iPhone 12’s launch by a month. That’s going to have an impact on the bottom line of your quarterly earnings. The company saw a 20% drop for the quarter, year-over-year. That’s hugely significant, causing the company’s stock to drop more than 4% in extended trading.
Apple’s diverse portfolio helped curb some of those revenue slides. While the pandemic has generally had a profound impact on consumer spending on “non-essentials,” changing where and how we work has helped bolster Mac and iPad sales, which were up 28% and 46%, respectively, year-over-year. It wasn’t enough to completely stop the iPhone stumble, but it certainly brings the importance of a diverse hardware portfolio into sharp relief.
China was a big issue for the company this time around — and the lack of a new, 5G-enabled iPhone was a big contributor. In greater China (including Taiwan and Hong Kong), the company saw a 28% drop in sales. There are a number of reasons to be hopeful about iPhone sales in Q4, however.
As I noted this morning, smartphone shipments were down almost across the board in China for Q3, per new figures from Canalys. Much of that can be chalked up to Huawei’s ongoing issues with the U.S. government. Long the dominant manufacturer in mainland China, the company has been hamstrung by, among other things, a ban on access to Android and other U.S.-made technologies. Apple’s numbers remained relatively steady compared to the competition and Huawei’s issues could present a big hole in the market. With 5G on its side, this next quarter could prove a banner year for the company.
Read Full Article
Google reveals a new Windows zero-day bug it says is under active attack
Google has dropped details of a previously undisclosed vulnerability in Windows, which it says hackers are actively exploiting. As a result, Google gave Microsoft just a week to fix the vulnerability. That deadline came and went, and Google published details of the vulnerability this afternoon.
The vulnerability has no name but is labeled CVE-2020-17087, and affects at least Windows 7 and Windows 10.
Google’s Project Zero, the elite group of security bug hunters which made the discovery, said the bug allows an attacker to escalate their level of user access in Windows. Attackers are using the Windows vulnerability in conjunction with a separate bug in Chrome, which Google disclosed and fixed last week. This new bug allows an attacker to escape Chrome’s sandbox, normally isolated from other apps, and run malware on the operating system.
In a tweet, Project Zero’s technical lead Ben Hawkes said Microsoft plans to issue a patch on November 10.
Microsoft didn’t independently confirm this date when asked, but said in a statement: “Microsoft has a customer commitment to investigate reported security issues and update impacted devices to protect customers. While we work to meet all researchers’ deadlines for disclosures, including short-term deadlines like in this scenario, developing a security update is a balance between timeliness and quality, and our ultimate goal is to help ensure maximum customer protection with minimal customer disruption.”
But it’s unclear who the attackers are or their motives. Google’s director of threat intelligence Shane Huntley said that the attacks were “targeted” and not related to the U.S. election.
A Microsoft spokesperson also added that the reported attack is “very limited and targeted in nature, and we have seen no evidence to indicate widespread usage.”
It’s the latest in a list of major flaws affecting Windows this year. Microsoft said in January that the National Security Agency helped find a cryptographic bug in Windows 10, though there was no evidence of exploitation. But in June and September, Homeland Security issued alerts over two “critical” Windows bugs — one which had the ability to spread across the internet, and the other could have gained complete access to an entire Windows network.
Updated with comment from Microsoft.
Read Full Article
Q3 earnings find Apple and Google looking to the future for hardware rebounds
“5G is a once-in-a-decade kind of opportunity,” Tim Cook told the media during the Q&A portion of Apple’s Q3 earnings call. “And we could not be more excited to hit the market exactly when we did.”
The truth of the matter is its timing was a mixed bag. Apple was, by some accounts, late to 5G. By the time the company finally announced that it was adding the technology across its lineup of iPhone 12 variants, much of its competition had already beat the company to the punch. Of course, that’s not a huge surprise. Apple’s strategy is rarely a rush to be first.
5G networks are only really starting to come into their own now. Even today, there are still wide swaths of users who will have to default to an LTE connection the majority of the time they use their handsets. The arrival of 5G on the iPhone was really as much about future proofing this year’s models as anything. Consumers are holding onto phones longer, and in the three or four years before it’s time for another upgrade, the 5G maps will look very different.
Clearly, the new iPhone didn’t hit the market exactly when Apple had hoped; the pandemic saw to that. Manufacturing bottlenecks in Asia delayed the iPhone 12’s launch by a month. That’s going to have an impact on the bottomline of your quarterly earnings. The company saw a 20% drop for the quarter, year-over-year. That’s hugely significant, causing the company’s stock to drop more than 4% in extended trudging.
Apple’s diverse portfolio helped curb some of those revenue slides. While the pandemic has generally had a profound impact on consumer spending on “non-essentials,” changing where and how we work has helped bolster Mac and iPad sales, which were up 28 and 46% respectively, year-over-year. It wasn’t enough to completely stop the iPhone stumble, but it certainly brings the importance of a diverse hardware portfolio into sharp relief.
China was a big issue for the company this time around — and the lack of a new, 5G-enabled iPhone was a big contributor. In greater China (including Taiwan and Hong Kong), the company saw a 28% drop in sales. There are a number of reasons to be hopeful about iPhone sales in Q4, however.
As I noted this morning, smartphone shipments were down almost across the board in China for Q3, per new figures from Canalys. Much of that can be chalked up to Huawei’s ongoing issues with the U.S. government. Long the dominant manufacturer in mainland China, the company has been hamstrung by, among other things, a ban on access to Android and other U.S.-made technologies. Apple’s numbers remained relatively steady compared to the competition and Huawei’s issues could present a big hole in the market. With 5G on its side, this next quarter could prove a banner year for the company.
Read Full Article
What crows teach us about death | Dr. Kaeli Swift
What crows teach us about death | Dr. Kaeli Swift
Rituals for the dead span much of the natural world, seen in practices from humans and elephants to bees, dolphins and beyond. With charm and playful insight, animal behaviorist Kaeli Swift delves into the life (and death) habits of crows and shares what their responses could reveal about our own relationship to mortality.
https://ift.tt/34GUESJ
Click this link to view the TED Talk
Rituals for the dead span much of the natural world, seen in practices from humans and elephants to bees, dolphins and beyond. With charm and playful insight, animal behaviorist Kaeli Swift delves into the life (and death) habits of crows and shares what their responses could reveal about our own relationship to mortality.
https://ift.tt/34GUESJ
Click this link to view the TED Talk
Facebook is limiting distribution of ‘save our children’ hashtag over QAnon ties
Facebook today confirmed that it will be limiting the distribution of the hashtag “save our children.” Over the past several months, the phrase — and ones like it — have become associated with QAnon. These terms have served to provide a kind of innocuous cover for the popular online conspiracy theory.
A spokesperson for the social network confirmed the move today, noting that child safety resources will be prioritized in search above those potentially tied to QAnon.
“Earlier this week, we stepped up how we enforce our rules against QAnon on pages, events, and groups,” a spokesperson told TechCrunch. “Starting today, we’re limiting the distribution of the ‘save our children’ hashtag given we’ve found that content tied to it is now associated with QAnon. When people search for it, they will now see the credible child safety resources.”
The company finally took action to remove the constellation of dangerous conspiracy theories with a ban on QAnon content across both Facebook and Instagram. It had previously announced a ban on QAnon groups that “discussed potential violence” but the expanded ban evinced a deeper understanding of how conspiracies draw in and radicalize regular users. The ban has actually proven quite successful so far, making it more more difficult for QAnon-related posts and accounts to be discovered and amplified.
Over the summer, the service began to crack down on QAnon-adjacent hashtags like SaveTheChildren. It even went so far as temporarily blocking the phrase, which, for around a century, has been associated with nonprofit youth organizations. “We temporarily blocked the hashtag as it was surfacing low-quality content,” Facebook told the press at the time. “The hashtag has since been restored, and we will continue to monitor for content that violates our community standards.”
By then, however, the movement had already gained life beyond social media, with several well-attended rallies being held across the U.S. and in different locations across the globe. Organizers have broadly purported to be protesting child exploitation, ranging from accusations of pedophilia among the Hollywood elite to outrage over the Netflix film “Cuties.”
In August, the U.S.-based Save the Children Federation, Inc. released a statement seeking to clarify and distance itself from the trend. “Our name in hashtag form has been experiencing unusually high volumes and causing confusion among our supporters and the general public,” the org wrote. “In the United States, Save the Children is the sole owner of the registered trademark ‘Save the Children.’ While people may choose to use our organization’s name as a hashtag to make their point on different issues, we are not affiliated or associated with any of these campaigns.”
Facebook’s crackdown on QAnon and adjacent #SaveTheChildren content come after the company allowed the dangerous conspiracy theory group to thrive on its platform for years, moving from the fringes of online life into its center. While President Trump and a handful of QAnon-friendly Republican political figures have given the conspiracies a boost, mainstream social networks allowed adherents to ferry the revelations of so-called “Q drops” from the obscure and often extreme message board 8chan into the center of American political life.
Some users happen upon conspiracy content organically, but algorithmic recommendations on platforms like Facebook and YouTube are known to usher users from the edges of conspiracies like QAnon into their often more extreme core ideas. Dedicated QAnon believers are responsible for a number of real-world violent actions, including an armed occupation of the Hoover Dam. Matthew Wright, the man who pled guilty to a terrorism charge for blocking the bridge, explained in a video that his agitation stemmed from President Trump’s failure to arrest his political enemies, which disappointed QAnon believers. Last year, a 29-year-old QAnon adherent shot and killed a mob boss who he believed was part of the “deep state” — a frequent preoccupation of Q followers.
Read Full Article
30 October 2020
Background Features in Google Meet, powered by Web ML
Video conferencing is becoming ever more critical in people's work and personal lives. Improving that experience with privacy enhancements or fun visual touches can help center our focus on the meeting itself. As part of this goal, we recently announced ways to blur and replace your background in Google Meet, which use machine learning (ML) to better highlight participants regardless of their surroundings. Whereas other solutions require installing additional software, Meet’s features are powered by cutting-edge web ML technologies built with MediaPipe that work directly in your browser — no extra steps necessary. One key goal in developing these features was to provide real-time, in-browser performance on almost all modern devices, which we accomplished by combining efficient on-device ML models, WebGL-based rendering, and web-based ML inference via XNNPACK and TFLite.
 |
| Background blur and background replacement, powered by MediaPipe on the web. |
Overview of Our Web ML Solution
The new features in Meet are developed with MediaPipe, Google's open source framework for cross-platform customizable ML solutions for live and streaming media, which also powers ML solutions like on-device real-time hand, iris and body pose tracking.
A core need for any on-device solution is to achieve high performance. To accomplish this, MediaPipe’s web pipeline leverages WebAssembly, a low-level binary code format designed specifically for web browsers that improves speed for compute-heavy tasks. At runtime, the browser converts WebAssembly instructions into native machine code that executes much faster than traditional JavaScript code. In addition, Chrome 84 recently introduced support for WebAssembly SIMD, which processes multiple data points using a single set of instructions, resulting in a performance boost of more than 2x.
Our solution first processes each video frame by segmenting a user from their background (more about our segmentation model later in the post) utilizing ML inference to compute a low resolution mask. Optionally, we further refine the mask to align it with the image boundaries. The mask is then used to render the video output via WebGL2, with the background blurred or replaced.
 |
| WebML Pipeline: All compute-heavy operations are implemented in C++/OpenGL and run within the browser via WebAssembly. |
In the current version, model inference is executed on the client’s CPU for low power consumption and widest device coverage. To achieve real-time performance, we designed efficient ML models with inference accelerated by the XNNPACK library, the first inference engine specifically designed for the novel WebAssembly SIMD specification. Accelerated by XNNPACK and SIMD, the segmentation model can run in real-time on the web.
Enabled by MediaPipe's flexible configuration, the background blur/replace solution adapts its processing based on device capability. On high-end devices it runs the full pipeline to deliver the highest visual quality, whereas on low-end devices it continues to perform at speed by switching to compute-light ML models and bypassing the mask refinement.
Segmentation Model
On-device ML models need to be ultra lightweight for fast inference, low power consumption, and small download size. For models running in the browser, the input resolution greatly affects the number of floating-point operations (FLOPs) necessary to process each frame, and therefore needs to be small as well. We downsample the image to a smaller size before feeding it to the model. Recovering a segmentation mask as fine as possible from a low-resolution image adds to the challenges of model design.
The overall segmentation network has a symmetric structure with respect to encoding and decoding, while the decoder blocks (light green) also share a symmetric layer structure with the encoder blocks (light blue). Specifically, channel-wise attention with global average pooling is applied in both encoder and decoder blocks, which is friendly to efficient CPU inference.
 |
| Model architecture with MobileNetV3 encoder (light blue), and a symmetric decoder (light green). |
We modified MobileNetV3-small as the encoder, which has been tuned by network architecture search for the best performance with low resource requirements. To reduce the model size by 50%, we exported our model to TFLite using float16 quantization, resulting in a slight loss in weight precision but with no noticeable effect on quality. The resulting model has 193K parameters and is only 400KB in size.
Rendering Effects
Once segmentation is complete, we use OpenGL shaders for video processing and effect rendering, where the challenge is to render efficiently without introducing artifacts. In the refinement stage, we apply a joint bilateral filter to smooth the low resolution mask.
 |
| Rendering effects with artifacts reduced. Left: Joint bilateral filter smooths the segmentation mask. Middle: Separable filters remove halo artifacts in background blur. Right: Light wrapping in background replace. |
The blur shader simulates a bokeh effect by adjusting the blur strength at each pixel proportionally to the segmentation mask values, similar to the circle-of-confusion (CoC) in optics. Pixels are weighted by their CoC radii, so that foreground pixels will not bleed into the background. We implemented separable filters for the weighted blur, instead of the popular Gaussian pyramid, as it removes halo artifacts surrounding the person. The blur is performed at a low resolution for efficiency, and blended with the input frame at the original resolution.
 |
| Background blur examples. |
For background replacement, we adopt a compositing technique, known as light wrapping, for blending segmented persons and customized background images. Light wrapping helps soften segmentation edges by allowing background light to spill over onto foreground elements, making the compositing more immersive. It also helps minimize halo artifacts when there is a large contrast between the foreground and the replaced background.
 |
| Background replacement examples. |
Performance
To optimize the experience for different devices, we provide model variants at multiple input sizes (i.e., 256x144 and 160x96 in the current release), automatically selecting the best according to available hardware resources.
We evaluated the speed of model inference and the end-to-end pipeline on two common devices: MacBook Pro 18 with 2.2 GHz 6-Core Intel Core i7, and Acer Chromebook 11 with Intel Celeron N3060. For 720p input, the MacBook Pro can run the higher-quality model at 120 FPS and the end-to-end pipeline at 70 FPS, while the Chromebook runs inference at 62 FPS with the lower-quality model and 33 FPS end-to-end.
| Model | FLOPs | Device | Model Inference | Pipeline |
| 256x144 | 64M | MacBook Pro 18 | 8.3ms (120 FPS) | 14.3ms (70 FPS) |
| 160x96 | 27M | Acer Chromebook 11 | 16.1ms (62 FPS) | 30ms (33 FPS) |
| Model inference speed and end-to-end pipeline on high-end (MacBook Pro) and low-end (Chromebook) laptops. |
For quantitative evaluation of model accuracy, we adopt the popular metrics of intersection-over-union (IOU) and boundary F-measure. Both models achieve high quality, especially for having such a lightweight network:
| Model | IOU | Boundary F-measure |
| 256x144 | 93.58% | 0.9024 |
| 160x96 | 90.79% | 0.8542 |
| Evaluation of model accuracy, measured by IOU and boundary F-score. |
We also release the accompanying Model Card for our segmentation models, which details our fairness evaluations. Our evaluation data contains images from 17 geographical subregions of the globe, with annotations for skin tone and gender. Our analysis shows that the model is consistent in its performance across the various regions, skin-tones, and genders, with only small deviations in IOU metrics.
Conclusion
We introduced a new in-browser ML solution for blurring and replacing your background in Google Meet. With this, ML models and OpenGL shaders can run efficiently on the web. The developed features achieve real-time performance with low power consumption, even on low-power devices.
Acknowledgments
Special thanks to the people who worked on this project, in particular Sebastian Jansson, Rikard Lundmark, Stephan Reiter, Fabian Bergmark, Ben Wagner, Stefan Holmer, Dan Gunnarson, Stéphane Hulaud and to all our team members who worked on the technology with us: Siargey Pisarchyk, Karthik Raveendran, Chris McClanahan, Marat Dukhan, Frank Barchard, Ming Guang Yong, Chuo-Ling Chang, Michael Hays, Camillo Lugaresi, Gregory Karpiak, Siarhei Kazakou, Matsvei Zhdanovich, and Matthias Grundmann.
How to foster productive and responsible debate | Ishan Bhabha
How to foster productive and responsible debate | Ishan Bhabha
The clash of ideas is fundamental to creativity and progress, but it can also be deeply destructive and create divisions within companies, communities and families. How do you foster productive debate while protecting against harmful speech and misinformation? Constitutional lawyer Ishan Bhabha lays out structures that organizations can use to navigate ideological disagreement and responsibly bring facts and context to a larger dialogue.
https://ift.tt/3ecGZGw
Click this link to view the TED Talk
The clash of ideas is fundamental to creativity and progress, but it can also be deeply destructive and create divisions within companies, communities and families. How do you foster productive debate while protecting against harmful speech and misinformation? Constitutional lawyer Ishan Bhabha lays out structures that organizations can use to navigate ideological disagreement and responsibly bring facts and context to a larger dialogue.
https://ift.tt/3ecGZGw
Click this link to view the TED Talk
Cloud infrastructure revenue grows 33% this quarter to almost $33B
The cloud infrastructure market kept growing at a brisk pace last quarter, as the pandemic continued to push more companies to the cloud with offices shut down in much of the world. This week the big three — Amazon, Microsoft and Google — all reported their numbers and as expected the news was good with Synergy Research reporting revenue growth of 33% year over year, up to almost $33 billion for the quarter.
Still, John Dinsdale, chief analyst at Synergy was a bit taken aback that the market continued to grow as much as it did. “While we were fully expecting continued strong growth in the market, the scale of the growth in Q3 was a little surprising,” he said in a statement.
He added, “Total revenues were up by $2.5 billion from the previous quarter causing the year-on-year growth rate to nudge upwards, which is unusual for such a large market. It is quite clear that COVID-19 has provided an added boost to a market that was already developing rapidly.”
Per usual Amazon led the way with $11.6 billion in revenue, up from $10.8 billion last quarter. That’s up 29% year over year. Amazon continues to exhibit slowing growth in the cloud market, but because of its market share lead of 33%, a rate that has held fairly steady for some time, the growth is less important than the eye-popping revenue it continues to generate, almost double its closest rival Microsoft.
Speaking of Microsoft, Azure revenue was up 48% year over year, also slowing some, but good enough for a strong second place with 18% market share. Using Synergy’s total quarterly number of $33 billion, Microsoft came in at $5.9 billion in revenue for the quarter, up from $5.2 billion last quarter.
Finally Google announced cloud revenue of $3.4 billion, but that number includes all of its cloud revenue including G Suite and other software. Synergy reported that this was good for 9% or $2.98 billion, up from $2.7 billion last quarter, good for third place.
Alibaba and IBM were tied for fourth with 5% or around $1.65 billion each.
Image Credits: Synergy Research
It’s worth noting that Canalys had similar numbers to Synergy with growth of 33% to $36.5 billion. They had the same market order with slightly different numbers with Amazon at 32%, Microsoft at 19% and Google at 7% and Alibaba in 4th place at 6%.
Canalys sees continued growth ahead, especially as hybrid cloud begins to merge with newer technologies like 5G and edge computing. “All three [providers] are collaborating with mobile operators to deploy their cloud stacks at the edge in the operators’ data centers. These are part of holistic initiatives to profit from 5G services among business customers, as well as transform the mobile operators’ IT infrastructure,” Canalysis analyst Blake Murray said in a statement.
While the pure growth continues to move steadily downward over time, this is expected in a market that’s maturing like cloud infrastructure, but as companies continue to shift workloads more rapidly to the cloud during the pandemic, and find new use cases like 5G and edge computing, the market could continue to generate substantial revenue well into the future.
Read Full Article
Smartphone shipments dip in China for Q3, led by Huawei decline
China was the first major global smartphone markets to rebound from the early days of the COVID-19 pandemic. Stringent lockdown measures were able to help the country recover from the virus relatively quickly during the first wave, as sales started to return well ahead of other areas.
In Q3, however, things have taken begun to decline again. New numbers from Canalys point to an 8% drop between quarters — and a 15% drop, year-over-year. The firm chalks much of the slow down to longtime market leader Huawei’s on-going issues with the U.S. government. The problems had a kind of cascading effect that served to impact the number two companies, Vivo and Oppo.
Image Credits: Canalys
“Huawei was forced to restrict its smartphone shipments following the August 17 US sanctions which caused a void in channels in Q3 that its peers were not equipped to fill. Huawei is facing its most serious challenge since taking the lead in 2016,” analyst Mo Jia said in a release. “If the position of the US administration does not change, Huawei will attempt to pivot its business strategy, to focus on building the [Harmony] OS and software ecosystem, as the Chinese government is eager to nurture home-grown alternatives to global platforms.”
Huawei dropped 18% in Mainland China, year-over-year. Vivo and Oppo posted similar declines at 13 and 18%, respectively. Xiaomi was able to make up ground at third place, gaining 19% y-o-y per the figures. Apple, meanwhile, remained relatively stead, in spite of the delated launch of the iPhone 12. Huawei’s continued struggles could provide a vacuum for the competition to fill.
Analyst Nicole Peng notes that the arrival of the 5G handset put the U.S. company in a strong position, looking forward, “iPhone 12 series will be a game changer for Apple in Mainland China. As most smartphones in China are now 5G-capable, Apple is closing a critical gap, and pent-up demand for its new 5G-enabled family will be strong.”
Read Full Article
Things to Keep in Mind When Deciding on an Online Casino
Online casinos have to measure themselves against each other within an increasingly competitive market. Everyone who has browsed the web for a casino has noticed that there are abundantly many to choose between. Thus, it can be challenging to know which one to pick and what separates them from each other. Let’s take a quick […]
The post Things to Keep in Mind When Deciding on an Online Casino appeared first on ALL TECH BUZZ.
WhatsApp is now delivering roughly 100 billion messages a day
WhatsApp, the popular instant messaging app owned by Facebook, is now delivering roughly 100 billion messages a day, the company’s chief executive Mark Zuckerberg said at the quarterly earnings call Thursday.
For some perspective, users exchanged 100 billion messages on WhatsApp last New Year’s Eve. That is the day when WhatsApp tops its engagement figures, and as many of you may remember, also the time when the service customarily suffered glitches in the past years. (No outage on last New Year’s Eve!)
At this point, WhatsApp is just competing with itself. Facebook Messenger and WhatsApp together were used to exchange 60 billion messages a day as of early 2016. Apple chief executive Tim Cook said in May that iMessage and FaceTime were seeing record usage, but did not share specific figures. The last time Apple did share the figure, it was far behind WhatsApp’s then usage (podcast). WeChat, which has also amassed over 1 billion users, is behind in daily volume of messages, too.
In early 2014, WhatsApp was being used to exchange about 50 billion texts a day, its then chief executive Jan Koum revealed at an event.
At the time, WhatsApp had fewer than 500 million users. WhatsApp now has more than 2 billion users and at least in India, its largest market by users, its popularity surpasses those of every other smartphone app including the big blue app.
“This year we’ve all relied on messaging more than ever to keep up with our loved ones and get business done,” tweeted Will Cathcart, head of WhatsApp.
Sadly, that’s all the update the company shared on WhatsApp today. Mystery continues for when WhatsApp expects to resume its payments service in Brazil, and when it plans to launch its payments in India, where it began testing the service in 2018. (It has already shared big plans around financial services in India, though.)
“We are proud that WhatsApp is able to deliver roughly 100B messages every day and we’re excited about the road ahead,” said Cathcart.
Read Full Article
Ad revenues and e-commerce boom boost Facebook earnings but US users down from COVID surge
Facebook reported its Q3 earnings today, including revenues of $21.5 billion, and net income of $7.8 billion. The company earned $2.71 in per-share profit during the three-month period.
Analysts had expected Facebook, the social giant, to earn a much-smaller $1.91 per-share off smaller revenues of $19.82 billion. The company also reported an average of 1.82 billion daily active users in September, up 12% compared to the year-ago period. Monthly actives were 2.74 billion, also up 12%. Both results were ahead of expectations.
Notably Facebook’s headcount rose sharply during the year, rising 32% compared to the year-ago period. That outstripped its 22% year-over-year revenue growth. The company’s total expenses rose 28%, again faster than its revenues.
Shares of Facebook are effectively flat in after-hours trading, up around 0.4% at the time of writing.
The company did not share a specific outlook for Q4 2020 or 2021 in its report, instead noting that it anticipates “fourth quarter 2020 year-over-year ad revenue growth rate to be higher than [its] reported third quarter 2020 rate,” along with stronger non-advertising revenues stemming from Oculus Quest 2 sales, the company’s new VR helmet.
Facebook did say that 2021 will bring a “significant amount of uncertainty.” A potential hurdle of Facebook will be the regulatory environment in Europe, and viability of transatlantic data transfers. Facebook says that its “closely monitoring the potential impact on our European operations as these developments’ progress.”
Analysts expect Facebook to generate revenues of $24.25 billion and per-share profit of $2.67 in the fourth quarter of 2020, and $100.0 billion in 2021 top line leading to $10.26 in per-share income.
What matters in all of this? That the core advertising market that seemed to bolster Snap’s own results has helped fill Facebook’s wings as well. Facebook noted in its earnings that it thinks that the “pandemic has contributed to an acceleration in the shift of commerce from offline to online,” leading to it experiencing “increasing demand for advertising as a result of this acceleration.” Twitter, meanwhile, saw ad revenue only marginally increase, about 8% from the year prior, as advertiser taste buds remain volatile.
That’s a tailwind from a secular shift. For Facebook, it could mean a good year’s growth.
It’s worth noting, however, that Facebook lost users in the U.S. and Canada — down to 196 million from 198 million last quarter — a decline that it attributed to a slowing surge from the abnormal highs seen in the midst of the lockdowns associated with the COVID-19 pandemic. So tailwinds, but also a return to normal patterns. And it expects this flat or down trend to continue into Q3, noting that “in the fourth quarter of 2020, we expect this trend to continue and that the number of DAUs and MAUs in the US & Canada will be flat or slightly down compared to the third quarter of 2020.”
Read Full Article
Daily Crunch: Google had a good quarter
Google releases its latest earnings report, Spotify is getting ready to raise prices and Excel gets friendlier to custom data types. This is your Daily Crunch for October 29, 2020.
The big story: Google had a good quarter
Google’s parent company Alphabet released its third-quarter earnings report this afternoon, coming in well ahead of Wall Street expectations thanks in large part to YouTube, which saw revenue rise to $5.0 billion (compared to $3.8 billion during Q3 2019).
Google Cloud also grew revenue from $2.4 billion last year to $3.44 billion in the most recent quarter. Overall, Alphabet reported revenue of $46.2 billion and earnings per share of $16.40, compared to analyst predictions of $42.88 billion in revenue and EPS of $11.21.
The company’s shares quickly rose 8.5% in after-hours trading.
The tech giants
Spotify CEO says company will ‘further expand price increases’ — Although the company didn’t detail its plans, CEO Daniel Ek said the hikes will take place in markets that are more mature for Spotify.
Microsoft now lets you bring your own data types to Excel — That means you can have a “customer” data type, for example, bringing in rich customer data from a third-party service into Excel.
Why Apple’s Q4 earnings look different this year — With Apple’s latest iPhone launch running a few weeks behind this year, it missed the window to be included on Q4.
Startups, funding and venture capital
Donut launches Watercooler, an easy way to socialize online with co-workers — The startup also announced that it has raised $12 million in total funding, led by Accel.
One-click housing startup Atmos raises another $4M from Khosla, real estate strategics and TikTok star Josh Richards — According to CEO Nick Donahue, users have started designing the “first dozen homes” on the platform.
Commissary Club wants to help formerly incarcerated people find community — While 70 Million Jobs focuses on helping people with criminal records find jobs, its new network Commissary Club is designed to be a place for folks to find community.
Advice and analysis from Extra Crunch
VCs poured capital into European startups in Q3, but early-stage dealmaking appeared to suffer — The VC trends of later and larger continue to change the landscape of private capital.
In the ‘buy now, pay later’ wars, PayPal is primed for dominance — Button’s Stephen Milbank writes that the greatest limitation to buy-now-pay-later adoption is its availability.
Twitter’s API access changes are chasing away third-party developers — On August 12, Twitter launched a complete rebuild of its 2012 API.
(Reminder: Extra Crunch is our membership program, which aims to democratize information about startups. You can sign up here.)
Everything else
Europe to limit how big tech can push its own services and use third-party data — Commission EVP Margrethe Vestager confirmed that a legislative proposal due in a few weeks will aim to ban what she called “unfair self-preferencing.”
Comcast says Peacock has nearly 22M sign-ups — But it’s not clear how many of them are paid versus free.
Tech optimism…in this economy? — The latest episode of Equity looks at big startup opportunities for the coming decade.
The Daily Crunch is TechCrunch’s roundup of our biggest and most important stories. If you’d like to get this delivered to your inbox every day at around 3pm Pacific, you can subscribe here.
Read Full Article
Twitter revenue rises 14%, but user growth fails to impress
Twitter continued to see its total traffic rise in the third quarter, thanks to that trifecta of returning sports, the presidential campaign and the COVID-19 pandemic. But there wasn’t nearly enough growth to appease Wall Street.
Twitter beat out analyst expectations on revenue and net income; However, Wall Street was stuck on Twitter’s user user figures, which showed minimal growth and sent shares lower in after-market trading. Twitter’s MDAUs — the company’s internal audience metric that measures monetizeable daily active users — hit 187 million in the third quarter. That’s a razor thin improvement from the 186 million the company reported in second quarter of this year, although it did represent a 29% rise from the 145 million in the same period last year. Analysts from FactSet had expected 195 million MDAUs.
That mDAU “growth” heads into flat-like-the prairie states territory when focused on the U.S. figures. The average US mDAU was 36 million for the third quarter, the same figure in the second quarter. In short, U.S. mDAUs are flat, flat, flat. It did grow from 30 million mDAUs in the third quarter of 2019. Meanwhile, average international mDAU was 152 million for the third quarter, compared to 115 million in the same period of the previous year and 150 million in the previous quarter.
Shares were down nearly 15% in after-market trading.
Twitter reported Thursday net income of $29 million in the third quarter, or 4 cents per diluted share, a decline from the same time period last year, when the company brought in a net income of $47 million at 5 cents per diluted share. Adjusted earnings were 19 cents a share.
The company’s revenue came in at $936 million, up 14% from the same period last year and 37% from the second quarter. Analysts had expected revenue of $777 million.
Twitter’s ad revenue also grew 15% to $808 million. Total ad engagement rose 27% over the same period in 2019. The return of live events as well as increased and previously delayed product launches helped boost ad revenue, Twitter CFO Ned Segal said.
“We also made progress on our brand and direct response products, with updated ad formats, improved measurement, and better prediction. We remain confident that our larger audience, coupled with ongoing revenue product improvements, new events and product launches, and the positive advertiser response to the choices we’ve made as we have grown the service, can drive great outcomes over time,” he added.
The U.S., Twitter’s biggest market, accounted for $513 million in revenue, a 10% increase YoY.
However, Twitter warned that the holiday season and U.S. election could impact advertiser behavior.
Read Full Article
Subscribe to:
Posts (Atom)