Posted by Jan Keller, Security Technical Program Manager
June has become the month where we’re inviting thousands of security aficionados to put their skills to the test...
In 2018, 23,563 people submitted at least one flag on their hunt for the secret cake recipe in the Beginner’s Quest. While 330 teams competed for a place in the CTF Finals, the lucky 10 winning teams got a trip to London to play with fancy tools, solve mysterious videos and dine in Churchill’s old chambers.
This June, we will be hosting our fourth-annual Capture the Flag event. Teams of security researchers will again come together from all over the globe for one weekend to eat, sleep and breathe security puzzles and challenges - some of them working together around the clock to solve some of the toughest security challenges on the planet.
Up for grabs this year is $31,337.00 in prize money and the title of Google CTF Champion.
Ready? Here are the details:
The qualification round will take place online Sat/Sun June 22 and 23 2019
The top 10 teams will qualify for the onsite final (location and details coming soon)
Players from the Beginner's Quest can enter the draw for 10 tickets to witness the Google CTF finals
Whether you’re a seasoned CTF player or just curious about cyber security and ethical hacking, we want you to join us. If you’re just starting out, the “Beginner's Quest” is perfect for you. Sign up to learn skills, meet new friends in the security community and even watch the pros in action. See you there! For the latest announcements, see g.co/ctf, subscribe to our mailing list or follow us on @GoogleVRP.
Posted by Bingyi Cao and Tobias Weyand, Software Engineers, Google AI
Last year we released Google-Landmarks, the largest world-wide landmark recognition dataset available at that time. In order to foster advancements in research on instance-level recognition (recognizing specific instances of objects, e.g. distinguishing Niagara Falls from just any waterfall) and image retrieval (matching a specific object in an input image to all other instances of that object in a catalog of reference images), we also hosted two Kaggle challenges, Landmark Recognition 2018 and Landmark Retrieval 2018, in which more than 500 teams of researchers and machine learning (ML) enthusiasts participated. However, both instance recognition and image retrieval methods require ever larger datasets in both the number of images and the variety of landmarks in order to train better and more robust systems.
In support of this goal, this year we are releasing Google-Landmarks-v2, a completely new, even larger landmark recognition dataset that includes over 5 million images (2x that of the first release) of more than 200 thousand different landmarks (an increase of 7x). Due to the difference in scale, this dataset is much more diverse and creates even greater challenges for state-of-the-art instance recognition approaches. Based on this new dataset, we are also announcing two new Kaggle challenges—Landmark Recognition 2019 and Landmark Retrieval 2019—and releasing the source code and model for Detect-to-Retrieve, a novel image representation suitable for retrieval of specific object instances.
Heatmap of the landmark locations in Google-Landmarks-v2, which demonstrates the increase in the scale of the dataset and the improved geographic coverage compared to last year’s dataset.
Creating the Dataset A particular problem in preparing Google-Landmarks-v2 was the generation of instance labels for the landmarks represented, since it is virtually impossible for annotators to recognize all of the hundreds of thousands of landmarks that could potentially be present in a given photo. Our solution to this problem was to crowdsource the landmark labeling through the efforts of a world-spanning community of hobby photographers, each familiar with the landmarks in their region.
Another issue for research datasets is the requirement that images be shared freely and stored indefinitely, so that the dataset can be used to track the progress of research over a long period of time. As such, we sourced the Google-Landmarks-v2 images through Wikimedia Commons, capturing both world-famous and lesser-known, local landmarks while ensuring broad geographic coverage (thanks in part to Wiki Loves Monuments) and photos sourced from public institutions, including historical photographs that are valuable to test instance recognition over time.
The Kaggle Challenges The goal of the Landmark Recognition 2019 challenge is to recognize a landmark presented in a query image, while the goal of Landmark Retrieval 2019 is to find all images showing that landmark. The challenges include cash prizes totaling $50,000 and the winning teams will be invited to present their methods at the Second Landmark Recognition Workshop at CVPR 2019.
Open Sourcing our Model To foster research reproducibility and help push the field of instance recognition forward, we are also releasing open-source code for our new technique, called Detect-to-Retrieve (which will be presented as a paper in CVPR 2019). This new method leverages bounding boxes from an object detection model to give extra weight to image regions containing the class of interest, which significantly improves accuracy. The model we are releasing is trained on a subset of 86k images from the original Google-Landmarks dataset that were annotated with landmark bounding boxes. We are making these annotations available along with the original dataset here.
Acknowledgments The core contributors to this project are Andre Araujo, Bingyi Cao, Jack Sim and Tobias Weyand. We would like to thank our team members Daniel Kim, Emily Manoogian, Nicole Maffeo, and Hartwig Adam for their kind help. Thanks also to Marvin Teichmann and Menglong Zhu for their contribution to collecting the landmark bounding boxes and developing the Detect-to-Retrieve technique. We would like to thank Will Cukierski and Maggie Demkin for their help organizing the Kaggle challenge, Elan Hourticolon-Retzler, Yuan Gao, Qin Guo, Gang Huang, Yan Wang, Zhicheng Zheng for their help with data collection, Tsung-Yi Lin for his support with CVDF hosting, as well as our CVPR workshop co-organizers Bohyung Han, Shih-Fu Chang, Ondrej Chum, Torsten Sattler, Giorgos Tolias, and Xu Zhang. We have great appreciation for the Wikimedia Commons Community and their volunteer contributions to an invaluable photographic archive of the world’s cultural heritage. And finally, we’d like to thank the Common Visual Data Foundation for hosting the dataset.
On Monday July 8, 2019 the Ad Manager API's ReportService will stop supporting reports that mix Ad Exchange historical reporting fields with Historical reporting fields. This change will impact all API versions.
This change is being made to ensure all API reports return accurate data. The Query Tool has never supported creating reports that mix Ad Exchange historical and Historical data, and in some cases this may lead to unexpected results. This change brings the API in line with the Query Tool.
How can I tell if I am impacted?
This change impacts approximately 2% of all reports run with the Ad Manager API. If you only run saved queries that were created with the Query Tool, this change does not impact you.
If you use any of the following columns, your report will be considered an Ad Exchange historical report, and you may only use either Dimensions prefixed with AD_EXCHANGE or date Dimensions including DATE, WEEK, DAY, HOUR or MONTH_AND_YEAR:
To prevent your reports from failing, you must convert your report to be either an Ad Exchange historical report or a Historical report.
To convert the report to an Ad Exchange historical report:
Remove any Columns not in the list above.
Remove any Dimensions that are not prefixed with AD_EXCHANGE, with the exception of the following Dimensions:
DATE
WEEK
DAY
HOUR
MONTH_AND_YEAR
Remove any DimensionAttributes for dimensions you removed.
Remove any PQL filters that applied to fields you removed.
To convert the report to a Historical report:
Remove any of the Columns in the list above.
Remove any Dimensions that are prefixed with AD_EXCHANGE.
Remove any PQL filters that applied to fields you removed.
If you have questions about this change or are unsure if it impacts you, reach out to us on the Ad Manager API forums. We'll be happy to check if you're impacted and work with you to update your applications.
Spring was blooming as we welcomed thousands of people to our big annual event: Google Cloud Next ’19. We heard great stories about how customers are growing their businesses with cloud, and we planted the seeds for lots of new ideas and connections. Read on to catch up on Next and what else was new last month with Google Cloud.
We introduced our new cloud service. Our big news in April, introduced by Sundar on stage at Next ’19, was Anthos, our open cloud platform. Anthos is a way for application developers to cut out the extra time they currently have to spend writing applications. For example, a developer might currently write code for an app running in their data center, and then have to rewrite it to work on another cloud, or in Google Cloud. Anthos lets developers package their code once, then run it across different types of clouds, which will save a ton of time.
There’s a new way to stash data in the cloud. If you’ve ever had to track down an old email containing important information, you know how suddenly necessary that email becomes, even if you received it years ago. That’s the idea behind archive storage, which is a way for companies to store data for a long time, more cheaply than data they might need to access regularly. At Next ‘19, we announced a new type of cloud storage that will let companies store that data in a way that will let them access it right away if and when they need it. It’s like a digital storage unit, and could be particularly useful for companies that have to store data for legal or compliance reasons.
There are some new ways to crunch the numbers. One popular way to use Google Cloud is for analyzing data. BigQuery is one of our cloud-based tools for storing data (known as a data warehouse) that can quickly store and analyze huge amounts of data. Companies can learn from the data and get new information that would be slow or difficult to get from traditional, non-cloud data warehouses, helping them make decisions and work more efficiently. These tools are popular and growing—the volume of data analyzed in BigQuery grew by more than 300 percent in the past year. We recently added new features that make it easier to get a company’s data into the cloud, and then to clean it up (by taking out duplicate information, for example) and organize it.
We wished Gmail a happy birthday. Now that Gmail is 15, it lets you schedule when you send emails and reply to a Google Docs comment directly within an email thread (no more flipping back and forth). Plus, Smart Compose in Gmail is now available in Spanish, French, Italian and Portuguese.
122 other things happened at Next. So many new products, features, and customer stories came out of Next ‘19 that we made a list of all 122 announcements from the show. The list starts with new regions coming online—those are new data centers that can help users get faster cloud access—and runs down all kinds of techie news. Scroll down to see some cool customer stories, such as UPS using GCP to help efficiently deliver millions of packages per day.
That’s a wrap for April! We’ll see you next month.
Over the past couple days, we hosted our fifth annual North American Creator Summit, where we brought together over a hundred of our most influential creators and artists for inspirational conversations. It was candid, it was fun, and there was a lot of latte art. From burning questions posed to YouTube Leadership to meaningful discussions with peers, the creators and artists who joined us fostered a special kind of community this year. There was energy and excitement in the air that was palpable and spirited, which gave the rollerskating extravaganza extra pizzazz.
We caught up with a few creators right after Susan Wojcicki (CEO), Neal Mohan (CPO) and Robert Kyncl (CBO) spoke about how creators and artists are the very heart of YouTube.
“The Internet’s always going to be asking for more, but hearing them talk about it in person — their steps and plans, what they’re planning to do to fix it, and how many people are part of the team to work on very specific issue — it’s comforting to hear them say that and know that they’re on our side,” said Lily Hevesh, the domino artist behind Hevesh5.
Interviews have been condensed for clarity.
Sam Tsui performing at Creator Summit (Photo by Alexander Stein)
Sam Tsui is a singer-songwriter who’s been on YouTube since August 2011. YouTube: Favorite part so far? Sam Tsui: It’s incredible to have the face-to-face with YouTube, with Susan [Wojcicki], and all the people who make this platform possible. There’s a lot to learn and a lot to know. … It’s always so amazing that YouTube wants us to come out and hear what we think, give us a heads up on what’s coming and all that good stuff. … Between that and all the fun activities, it’s totally amazing, overwhelming, and a ton of fun. YouTube: Key takeaway after hearing Susan, Neal and Robert on stage? Sam Tsui: [It’s] wonderful to hear how healthy the platform is. The statistics about the number of creators who have over one million subs is growing, and the engagement, and the fact that this is, as ever, the place for the kind of stuff you want to be doing.
Natalie and Dennis got married in December 2017 after six years of dating. Natalie also has her own separate channel called Natalies Outlet. YouTube: Favorite part so far? Natalie: It’s such an honor to be in a room full of some of the most powerful people on the Internet. Especially having the speakers come in and talk to you so genuinely, and without a third wall and so real. Sometimes I think we get caught up in the business of YouTube, and it all comes down to the passion. Dennis: The hospitality is always so nice. We always have packages when we arrive. We feel cared for. YouTube: Key takeaway after hearing Susan, Neal and Robert on stage? Natalie: Sometimes you don’t really see what they’re doing behind-the-scenes. You just think, “Oh, they represent YouTube.” But they really did show that they’re working on policies. They’re working on making sure as creators, we’re continuing to monetize. It’s nice to see that they’re so caring and they answered real questions, even though [the questions are] kind of hot sometimes.
Hyunee’s mukbang channel has 1.2 million subscribers, and this is her first year at Creator Summit.
YouTube: Key takeaway after hearing Susan, Neal and Robert on stage? Hyunee Eats: We complain a lot about the faults that they have, but knowing that they’re working hard to improve everything and actually seeing them talk to us in person has helped us learn about what they actually do behind-the-scenes. ... They’re like real people, like us.
Lily Hevesh has been making domino art videos since 2009. This is also her first Creator Summit!
YouTube: Favorite part so far? Lily Hevesh: I don’t really get the opportunity to meet other people who make videos for a living, so bringing all the top creators in one room is super exciting. YouTube: Key takeaway after hearing Susan, Neal and Robert on stage? Lily Hevesh: Just seeing them in person and hearing them speak — to me, it felt like they really do deeply care about the creators, fans and advertisers. And they’re trying their best to please all of them. While there are lots of issues with the site, they’re working as hard as they can to try and solve them.
Remi Cruz is a 23-year-old lifestyle and wellness creator with 2.5 million subscribers.
YouTube: Favorite part so far? MissRemiAshten: We’re only on day one, and this has been my favorite one, for sure. Getting to see Julie Rice, the co-founder of SoulCycle, and getting to see her interact with Blogilates, who’s one of my favorite YouTubers and one of my really good friends. I feel like they’re two people I look up to so much, and I live by Soulcycle so I genuinely feel like they tailored a lot of stuff to our interests today. YouTube: Key takeaway after hearing Susan, Neal and Robert on stage? MissRemiAshten: There are so many things that go on behind the scenes that I don’t even know about. So it’s interesting to hear that. … It’s nice that YouTube has this whole conference for us in general, because no other platform does it.
Today, people want to watch videos that relate to their interests and passions – whenever they choose, on any device. For some of us, it’s guitar lessons, but for others, it’s baking their own masterpiece alongside experts like How to Cake It. Everyone’s primetime is personal, and what you watch often reflects who you are.
There’s no better place than YouTube to find the creators, music and shows that you love. And that’s why YouTube is #1 in reach and watchtime among ad-supported streaming media, according to Comscore.1
For advertisers, this means being able to reach people who are increasingly cutting the cord. In fact, YouTube now reaches more 18 to 49 year-olds in an average week than all cable TV networks combined.2
Tonight, at our eighth annual Brandcast event, we celebrated the creator, entertainment and music content that audiences love on YouTube. We also showcased new ways we can better help brands reach viewers in an on-demand world.
It’s all about the engagement and results
YouTube is focused on helping brands reach their audiences and drive business impact. To deliver this, we announced that Nielsen Total Ad Ratings (TAR) is now supported on YouTube. This lets brands compare YouTube and TV reach apples-to-apples.
Across the 20 Total Ad Ratings (TAR) studies we’ve commissioned with Nielsen for brands like Colgate and the U.S. Navy, YouTube has consistently driven incremental reach at a more balanced frequency compared to TV.3
And across all Marketing Mix Modeling studies we’ve commissioned from Nielsen in the past two years that measured Google Preferred, we had greater sales per impression than TV in every one of them.4
To give advertisers more opportunities to validate their investments, we announced that Nielsen Catalina Solutions (NCS) Sales Lift will be available for Google Preferred campaigns before the end of 2019. The tool helps brands measure the lift in offline sales for U.S. consumer packaged goods brands, and provide deeper performance insights across audiences, creative and more.
Karate-chopping the paywall: Bringing more YouTube Originals to advertisers
From Hollywood celebrities to top YouTube creators, YouTube Originals deliver the content audiences love. We saw this reflected in last year’s star-powered slate, which amassed over 2.5 billion views across 50 shows like Will Smith's "The Jump" or Liza Koshy’s "Liza on Demand."5
But some of our best content was still behind the paywall, including “Step Up: High Water,” which achieved unprecedented viewership for its season two premiere, and “Cobra Kai,” the critically acclaimed reboot of the 1984 classic “The Karate Kid.” In its rookie season, “Cobra Kai” received 11 award nominations (including an Emmy nomination), the Golden Tomato Award for Best TV Drama and a 100-percent rating on Rotten Tomatoes – making it the highest-reviewed reboot of all time. And just after its season two premiere last week, which reached 20 million views in six days, Parrot Analytics confirmed Cobra Kai is the most in-demand digital original series in the world right now.6
Today, we announced that all new YouTube Original series and specials will soon be available for fans around the world to watch for free with ads — just like they enjoy other content on the platform. And we are excited to confirm this includes the renewal of “Cobra Kai” for season three.
This gives advertisers more opportunities to engage with a broader audience, drive meaningful results and align with top Hollywood talent and YouTube creators.
Our new slate builds upon the success of our first two years. We’re continuing to focus on stories that delight us from our favorite personalities, inspire us through rich cultural moments and educate us on topics that matter. From returning hits like “Kevin Hart: What the Fit” and “Liza on Demand” to new shows — with Dude Perfect, Markiplier and a top secret project with Justin Bieber — the slate celebrates the diversity, authenticity and spirit characteristic of the YouTube community.
YouTube Originals will be available to advertisers through bespoke sponsorships or Google Preferred lineups.
More music live streams, more opportunities for brands
Last month, YouTube’s live stream of Coachella's first weekend was the most viewed yet, earning over 82 million live views — an increase of over 90 percent compared to 2018.7
After nine years of success in the desert, we’re doubling down on music festivals, starting with a new two-year partnership to live stream Lollapalooza this August. In addition to the curated live stream over the four-day festival, for the first time ever, fans will get additional custom content that gives them unique access to the festivals' artists.
These key moments in music, along with other cultural events like YouTube FanFest, provide more opportunities to reach audiences with the content they are most passionate about.
To help advertisers better reach their audience, we’re evolving our Google Preferred offering to reflect changing viewing habits, and help advertisers connect with those viewers where they are.
Our proprietary algorithm, the P-Score, looks at the popularity and viewer passion of specific content — things like the amount of repeat views and how often videos are shared to keep Google Preferred Lineups fresh. It also draws on protection, which places a greater emphasis on content suitability where Google Preferred Lineup ads will only serve on videos that are reviewed first by machine classifiers, then verified by humans.
Tonight, we shared that we’re adding two new ways to help the P-Score be even more useful to brands. There’s platform (for videos frequently watched on larger screens) and production (for content that has higher production values like advanced camera work or sound quality) to bring more high-quality inventory into Google Preferred.
Where our TV offering is going and how it’s growing
Our TV offering comes in two parts: The first is YouTube on TV screens, which is the YouTube app optimized for Internet-connected TVs. Tonight, we shared that connected TVs represent our fastest growing screen, where watch time now tops 250 million hours per day.8
The second is YouTube TV, which launched two years ago to give people a new and reimagined way to enjoy cable-free live TV along with the best of YouTube content. Now with over 70 broadcast and cable channels, YouTube TV is available nationwide and will be accessible as a standalone lineup in Google Preferred this upfront season. This change, along with audience guarantees and show-level exclusions, allows advertisers to zero in on live and on-demand inventory. And most importantly, brands can go beyond demographics to reach audiences based on their interests – just like with the core Google Preferred offerings.
For today's viewers, primetime is personal. With this expanded offering, advertisers and brands can reach and influence audiences on proven platforms like YouTube.
Check out #Brandcast to see highlights from the night.
1 Comscore OTT Intelligence and Custom Reporting based on Total Home Panel, September 2018, U.S
2 Google commissioned Nielsen study, U.S., May 2018. People age 18-49. Average weekly reach for Cable TV and YouTube (includes mobile, desktop, and living room devices)
3 "Nielsen Total Ad Ratings (TAR) Meta Analysis, including all 20 TAR US studies inclusive of YouTube in-app traffic commissioned by Google running from January 2018 to March 2019 across desktop, mobile and TV. Incremental Reach Value calculated as YouTube’s on-target incremental reach / YouTube’s Total on-target Reach. Standard Deviation 22%. On Target Frequency calculated as On Target Impressions / On Target Reach. Standard Deviation 6.5
4 Commissioned Nielsen MMM Meta-Analysis, US 2017-2018. Base: all the studies Google commissioned in 2017 and 2018 where Google Preferred Lineups was measurable, a total of 20 MMM across 3 categories. Count based on incremental offline retail sales measured as two-tailed significance > 90%. Effectiveness is defined as incremental sales per impression. Incremental offline retail sales measured for TV represents average across all TV broadcasters. calculated as On Target Impressions / On Target Reach. Standard Deviation 6.5
5 YouTube Originals YouTube Analytics, Global, Jan 2018-Dec 2018
8 YouTube Internal Data, Global, March 2019. Based on 90 day average of WatchTime for Living Room Devices, which include smart TVs, Roku/Apple TV, and game consoles
Earlier this month, my teammates and I attended the 2019 Net Inclusion conference in Charlotte, North Carolina. As the Government and Community Affairs team (formerly known as the “Community Impact Team”) for Google Fiber, we are responsible for building and investing in partnerships that help narrow the digital divide across the Google Fiber footprint.
The gathering, organized by the National Digital Inclusion Alliance (NDIA), provided a valuable opportunity to listen and learn from digital inclusion leaders on the front lines of improving their respective communities. We also had the chance to see the progress of the broader digital inclusion movement -- a movement that is bigger than any one program, strategy or organization.
As a Google Fiber city and home to several leading digital inclusion organizations, the City of Charlotte offered many learning opportunities for those interested in starting, improving or expanding digital inclusion programming for residents. Many of these learnings are captured in the recently published Digital Inclusion Start-Up Manual, which was authored by NDIA and sponsored by Google Fiber. The manual includes recommendations for developing a community digital inclusion program, among other resources.
Throughout the conference, dozens of digital inclusion leaders took part in lightning round sessions featuring innovative programs from across the country. From leveraging technology to expand educational opportunities across the world to the development of low cost digital literacy evaluation tools, our team was inspired by these innovative approaches from community-based practitioners. In fact, it was during the lightning rounds of last year’s conference that we first discovered Libraries Without Borders’ Wash and Learn Initiative. Since then, we have kicked off a partnership with Libraries Without Borders to pilot their program in San Antonio.
We were also excited to see the announcement of the 2019 Charles Benton Digital Equity Champions. Honorees included Casey Sorensen, CEO of Minnesota-based PCs for People, and Munirih Jester, Digital Inclusion Program Manager for the San Antonio Housing Authority. Munirih began her role as one of the very first NTEN Digital Inclusion Fellows, a program Google Fiber co-founded with NTEN in 2015 and that is now in its fifth cohort. These champions manage programs aimed at reducing the digital divide within their respective communities.
Many thanks to the NDIA team for organizing another successful event, the Charlotte Google Fiber team for hosting a fun-filled reception, and local digital inclusion leaders for opening your doors to Net Inclusion visitors.
We know the digital divide remains a persistent challenge — and that none of us can solve it alone. Google Fiber has learned a lot about digital inclusion each year since our launch in 2012; and the conference made clear that the digital inclusion ecosystem continues to grow and get stronger. We continue to learn from leaders across the country working hard to get more of their neighbors connected, and we look forward to chipping away at the divide right alongside them.
Munirih Jester, Digital Inclusion Program Manager at the San Antonio Housing Authority, receives the 2019 Charles Benton Digital Equity Champion award from Adrianne Furniss, Executive Director of the Benton Foundation
Posted by Clarissa Ramon, Government and Community Affairs Manager, San Antonio
Hi everyone! We've just released Chrome Beta 75 (75.0.3770.16) for Android: it's now available on Google Play.
You can see a partial list of the changes in the Git log. For details on new features, check out the Chromium blog, and for details on web platform updates, check here.
If you find a new issue, please let us know by filing a bug.
The Chrome team is excited to announce the promotion of Chrome 75 to the beta channel for Windows, Mac and Linux. Chrome 75.0.3770.18 contains our usual under-the-hood performance and stability tweaks, but there are also some cool new features to explore - please head to the Chromium blog to learn more!
A full list of changes in this build is available in the log. Interested in switching release channels? Find out how here. If you find a new issue, please let us know by filing a bug. The community help forum is also a great place to reach out for help or learn about common issues.
- Chris Seeley, Ad Manager API Team
