has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

I want to know more about the newly announced Saudi constellation! When will we see imagery? :)

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

Hey everyone. Shay is going to join us in just a few minutes. The AMA will start at 11am PST sharp and will end at noon, so we have a full hour of Q&As. Let?s try to keep the questions short and succinct and if you want to upvote an answer simply add a :upvote: reaction and vote it up.

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

Satellites FTW

has joined #ama-shay-03-2016

Hey Y?all! Thanks for having me!

Alright peeps. Let?s welcome @sharnoy to the channel. Shay comes to us from Digital Globe and is the VP and GM of Platform. In case you?re not aware of Digital Globe does - they build and shoot satellites into space and provide imagery of Earth and resell it to the public sector and private companies. Their goal is to provide real-time imagery and make it scalable and accessible to everyone. Welcome Shay! Thanks for doing this.

Definitely!

@willcadell: you can get plenty of imagery ? once the constellation launches

Welcome, thanks for doing this AMA

Looks like everyone is typing out questions lol

Biggest technical challenges?

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

@sharnoy: Who do you guys use to get your constellations into space?

(Granted, that's a pretty open question, have at 'er)

@arcvancouver: thats a huge question. Lots of challenges ? our data is really big (1900x the size of landsat by file size), machines are still pretty dumb (getting better with deep learning and machine learning)

@julien: we use fat rockets to get into space. Our latest satellite, Worldview-3, went up on an Atlas V rocket :rocket:

has joined #ama-shay-03-2016

@sharnoy: can you see me?

(ie what is the best resolution you currently have)

@arcvancouver: other tech challenges is how to make hard concepts around remote sensing and data analytics more accessible. Hiding the power is a must if we are to get broad adoption.

Have you met Elon Musk?

@sharnoy: What?s the goal with the imagery. Will it be accessible to the development community for a relatively cheap price or are you guys focusing on reselling it to major players slash countries, major corporations etc.

We see that DG is totally changing its business model with the new big data platform, can you tell us about how this platform is a big step forward Imagery as a Service instead of just selling pixel

@fil: 30cm is the best so we can?t identify you or anything like that but we can definitely see crowds

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

Any comments on Terra Bella getting into the game?

@themapsmith: Some of our folks at the company have met Elon Musk. Apparently he is a total baws. I just tweeted a great article about him @iheartcrowds

@sharnoy: Rad! I loved the mega-writeup by Wait But Why :rocket:

has joined #ama-shay-03-2016

@julien: My entire goal at DigitalGlobe is to make more of our data accessible for development. Its a non-trivial challenge because of the sheer volume (100 Petabytes!). In fact, if you look at my Finder window you will see about 10 slideshows titled something like ?Data_out_of_jail...'

has joined #ama-shay-03-2016

@vincents: A realization I had when I joined digitalglobe is that to enable large scale analytics, you need to solve the technical hurdles AND the business hurdles. People typically can?t pay x$ per km2 when doing large scale analytics over broad areas or long time series. Thats because the value may not scale linearly with km2.

can a casual user access imagery, say to get a picture of their house?

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

@talldave: Terra Bella is the same Skybox crew with the same Skybox vision. Its a small community and I wish them well. I have said this before in public events ? if all we do is take away the smidgen of market share away from each other, this industry is doomed. Its all about growing the pie. I totally mean that.

@fil mapbox satellite basemap use DG imagery, but 30cm resolution is not available everywhere

has joined #ama-shay-03-2016

@fil: Casual user has multiple meanings depending on sophistication. The most casual user can see their house on their mobile phone by looking at Google, Apple, Mapbox, HERE, Bing, or a number of other international players that I am not sure if we have announced publicly but its pretty obvious

@vincents: thats true. The world is REALLY big! Our constellation collects ~3million km2 a day across 4 satellites. The landmass of the earth is 150m km2!

@sharnoy: Have you guys had talks with the NSA or other security agencies regarding the impact of allowing almost anyone access to such precise satellite imagery with a high refresh rate and what do you guys do to mitigate the technology from getting in the wrong hands? Or is this even an issue we should be worrying about?

@vincents: Also, our satellites are aimed at the most interesting places of the earth to get the most relevant imagery.

has joined #ama-shay-03-2016

@sharnoy: can you tell us how and why you created tomnod

has joined #ama-shay-03-2016

@julien: Good question. I can?t speak about specific conversations. The US Government is our largest customer and the owner of our license to operate so we take the responsibilities of operating safely and securely very very seriously. :dark_sunglasses:

Hi, @sharnoy! :wave: I?m curious to hear any fresh news or thoughts about WorldView-4 (formerly GeoEye-2). You?re going to have quite the constellation of half-meter-and-better hardware. Scaling up coverage and cadence is great, but are there qualitative changes you see coming from WV4?

@vincents: We started Tomnod to scratch our own itch. Near the end of grad school we launched an expedition to search for the tomb of Genghis Khan in Mongolia. Through the standard turn of events, we got the GeoEye foundation to donate imagery, National Geographic to sponsor, and the Mongolian Academy of Sciences to support. All of a sudden we had Terabytes of imagery to comb through of northern mongolia searching for anomalies. How can you train a machine to look for something that you don?t know what it looks like? Thats how we came across the idea of crowdsourcing of imagery. http://exploration.nationalgeographic.com/

@vincents: Tomnod actually means Big Eye in Mongolian https://vimeo.com/28058031

(video live from the field during our second summer in Mongolia!)

@sharnoy: excellent

I suppose there might be some discussions about whose are ?the wrong hands?. Do you have a programme to work with investigative journalists ? ie help us use the system to find relevant data?

@vruba: one of the things that people don?t talk about is the collection capacity and agility of these satellites (I certainly didn?t think about it before joining DigitalGlobe). WV-4 is a collection machine with really solid agility to collect at various angles when clouds might be obscuring (see for example the shot we took of Kathmandu right after the earthquake at a high off-Nadir angle due to cloud cover).

I wrote a blog post about that shot :wink:

@vruba: ha! I was going to reference your blog post then :simple_smile:

has joined #ama-shay-03-2016

@vruba: Getting the shot is sweet but from an analytics point of view it makes automatic feature and change detection more challenging than just looking straight down.

for example, I suppose there is a way to query ?all gold mining fields in such an such region? and retrieve all images, sorted by year etc

but journos don?t normally have access to this kind of intelligence

has joined #ama-shay-03-2016

@vruba: Stupid atmosphere :face_with_rolling_eyes: ? at any given time, 70% of the earth is covered in clouds so looking straight down isn?t very productive for many applications.

@vruba: shoot, this is going to be published on the web?. For the record I love the atmosphere :partly_sunny_rain::rain_cloud::+1:

@fil: I like the way you are thinking. I can totally see you building an end user app on GBDX targeted at that use case.

@vruba?s blog post ??> https://www.mapbox.com/blog/nepal-imagery-collection/

@fil: We do get a lot of inbounds (I still read every tomnod crowd emails btw to stay connected with the world rather than just the geo community) and we have had multiple requests for assistance with investigations of missing persons, crimes, etc? Of course we help however we can. Interesting how these requests often come from our ?crowd-facing? presence rather than our ?geo-facing? presence

In other news, our team loaded up our metadata into an elastic search instance on our platform and have been able to whip up some pretty cool visualizations.

@sharnoy uploaded a file: https://gisdevs.slack.com/files/sharnoy/F0STJFD52/catalogagg.kmz

I?d be super interested at trying something along those lines (& I?d love to elaborate more on this issue, but now I?ve got to go, sorry sorry).

If the question backlog is clear, here?s a real softball for @sharnoy :wink: ? It looks like GBDX is going to be a true platform, with lots of different people running their projects on it. But is there any particular task or technique that you?re looking forward to seeing run on it? Any interesting capabilities you think people might overlook? Or any cool experiments you?ve tried already?

(GBDX, for anyone who hasn?t seen it: http://developer.digitalglobe.com/gbdx/)

@vruba: that is a softball!

@vruba: I think we are still at the infancy of feature detection and deep learning as an industry and as a technology sector. Im excited to ride that wave and see where it takes us.

@sharnoy: Are you guys hiring?

@vruba: a big question will be making remote sensing and machine analytics at scale easier to use and more accessible

@sharnoy: How is @bddavidson working out? :stuck_out_tongue_winking_eye:

@julien: yes! Technical roles and product roles working on the Platform side.

@themapsmith: Hey now. I don?t need my yearly review via slack :stuck_out_tongue:

@julien: jobs in Colorado and in DC/Herndon

@sharnoy: help out a newbie here: Is it possible for, say, a farmer to submit drone photography of their fields from yesterday to your system to infer important information? Even if your data is 3 months old at 30cm, it wouldn?t matter to farmers because it?s? well, 3 months old.

what algorithms and classification techniques are you guys using for machine learning?

e.g. Random Forest, SVM, etc

@andest01: we have been working with a couple of the leading drone companies. One of the added value we provide is providing historical context for comparisons. For example, if you go out and buy a drone today, you fly it and get some information about your field (eg certain NDVI values). That information is MUCH more valuable if you are able to compare it to the NDVI values from past seasons where you have ground truth as to what the farm output was.

@andest01: so people are absolutely bringing their own data and combining it with whatever digitalglobe data.

@sharnoy: Are there any developments or partnerships you?ve made to make relevant imagery more accessible to crisis mappers responding to events like the Nepal earthquake, Syrian refugee migrations, food security in Africa?

@sharnoy: thanks!

: We?ve got about 5 minutes left in case you have any questions.

@robodonut: We enabled on the platform many of the core remote sensing techniques by implementing the ENVI toolkit (e.g. band combos, LULC, etc?). In addition, our team is using Random Forest, SVMs with heuristic feature vectors for various applications, convolutional neural nets, and a bunch of other stuff. Our approach is to make our own DigitalGlobe algorithms available for you to use on the platform but also allow you to build your own and bring them to the data. Such that even if you are really really good at detecting coniferous trees and have a special built algorithm for that, you don?t have to also build your own atmospheric compensation algorithm, or cloud detector but you can just focus on your own task at hand.

@max: yes. a lot. We respond to events around the globe every week. Some of them are on the front page of the news and others aren?t but either way we are there. In all of them, or nearly all of them, our data makes its way to people responding on the ground. Often times we launch a tomnod campaign alongside it to make not only the imagery but also the derived information accessible.

@max: For the larger events where many people from around the world are sending resources we often make our imagery freely available for anyone (as was in the case of Haiyan or Nepal earthquake).

@max: Our main goal is to do what is right for the people on the ground and be outcome driven.

Awesome. Well it just hit noon. I think everyones going to be heading out to lunch real soon. But thank you for joining us Shay for the AMA. It?s been really educational and we hope that DG keeps killin it!

Thanks @sharnoy

Thanks for having me. If anyone is interested in a GBDX eval account. Drop me a message or sign up at http://developer.digitalglobe.com

Awesome

has joined #ama-shay-03-2016

Thanks for joining! Here is a link to my latest blog post with a signup for platform access at the bottom. http://www.digitalglobeblog.com/2016/02/22/helping-facebook-connect-the-world-with-deep-learning/

@ambell has left the channel

has joined #ama-shay-03-2016

@gisdev-km has left the channel

/points @sharnoy +1000

jeremy has awarded @sharnoy 100 points

/points @sharnoy 100 for a great AMA

robodonut has awarded @sharnoy 100 points

has joined #ama-shay-03-2016

@vruba has left the channel

@aaron has left the channel

@amit has left the channel

has joined #ama-shay-03-2016

has joined #ama-shay-03-2016

@fgcarto has left the channel

has joined #ama-shay-03-2016

@manny has left the channel

@dustin has left the channel

@stuporglue has left the channel

@jeremy has left the channel

has joined #ama-shay-03-2016

@evan has left the channel

@jvilleneuve has left the channel

@julien archived the channel (w/ 38 members)

has joined #ama-shay-03-2016