The UN Open GIS initiative's Spiral 4 for geospatial data collection - TIB AV-Portal

The UN Open GIS initiative's Spiral 4 for geospatial data collection

00:00

Formale Metadaten

Titel

The UN Open GIS initiative's Spiral 4 for geospatial data collection

Alternativer Titel

The UN Open GIS initiative – Vision, strategies, and achievements

Serientitel

FOSS4G Bucharest 2019

Anzahl der Teile

295

Autor

Brovelli, Maria

Eom, Kyoung-Soo

Kim, Byeong-Gon

Kannan. Anandavel

Mitwirkende

Lizenz

CC-Namensnennung 3.0 Deutschland:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen.

Identifikatoren

10.5446/43383 (DOI)

Herausgeber

Open Source Geospatial Foundation (OSGeo)

Erscheinungsjahr

Sprache

Inhaltliche Metadaten

Fachgebiet

Genre

Abstract

With an aim of developing and delivering open source geospatial solutions to the UN, the UN Open GIS Initiative was established in March 2016, taking full advantage of mission partner expertise from UN member states, technology contributing countries, international organizations, academia, NGO’s and private sector. In order to fulfil the diverse requirements of UN field operations, the scope of the UN Open GIS Initiative covers the geospatial software of the UN Spatial Data I nfrastructure. The activities are organized into four working groups, referred to as ‘spirals’; Spiral one for geo-portal and system infrastructure, Spiral two for capacity building, Spiral three for geo-analytic functions, and Spiral four for geospatial data collection. This talk will cover an overall introduction of the UN Open GIS initiative including the vision, strategies, technical roadmaps, and achievements of each working group. The experiences and lessons learned from the initiative will be shared during the talk, which we hope will be helpful to the UN as well as developing countries. We will also discuss how to leverage the FOSS4G community and the UN Open GIS Initiative.

Schlagwörter

Sprache

Text

Bild

00:00

TopologieMultiplikationsoperatorWeb SiteCoxeter-GruppeEreignishorizontBitOffene MengeGrundraumMinkowski-MetrikVorlesung/Konferenz

00:50

SpiraleOffene MengeAnalysisDatenverwaltungDienst <Informatik>Array <Informatik>Inhalt <Mathematik>VideokonferenzDigitale PhotographieFunktion <Mathematik>Content ManagementSichtenkonzeptElementargeometrieDatenfeldRichtungDickeFolge <Mathematik>Einbettung <Mathematik>Gerichtete MengeSpezialrechnerPrädikat <Logik>AbfrageTrajektorie <Kinematik>Tomcat <Programm>SystemplattformMultimediaPaarvergleichGEDCOMAppletBrowserMathematikProgrammbibliothekTemporale LogikREST <Informatik>Konfiguration <Informatik>ImplementierungVideokonferenzKonfiguration <Informatik>Bildgebendes VerfahrenOpen SourceDigitale PhotographieObjekt <Kategorie>VersionsverwaltungArray <Informatik>TesselationPrädikat <Logik>SoftwareBitMultimediaOrdnung <Mathematik>Elektronische PublikationResultanteTrajektorie <Kinematik>AbfrageProgrammbibliothekKonstruktor <Informatik>Physikalisches SystemInformationCASE <Informatik>RahmenproblemMetadatenAnalysisAdditionMigration <Informatik>DatenverwaltungOffene MengeSoftwaretestDatenfeldKonditionszahlFolge <Mathematik>Gemeinsamer Speichersinc-FunktionFamilie <Mathematik>SichtenkonzeptMAPElementargeometrieStichprobenumfangFunktionalPaarvergleichZweiPhasenumwandlungMultiplikationsoperatorEinfache GenauigkeitKartesische KoordinatenFundamentalsatz der AlgebraDreiecksfreier GraphComputerspielProdukt <Mathematik>ValiditätEchtzeitsystemPortal <Internet>DatenstrukturWinkelKnotenmengeStreaming <Kommunikationstechnik>SystemprogrammCoxeter-GruppeGrundraumSpider <Programm>Automatische HandlungsplanungVollständigkeitRuhmasseZentralisatorSummierbarkeitSchlussregelInformationsmanagerRechter WinkelTotal <Mathematik>System FProzess <Informatik>ClientHypermediaMinkowski-MetrikMapping <Computergraphik>Dienst <Informatik>Vollständiger VerbandNichtlinearer OperatorWeb SiteDruckspannungSchießverfahrenDemoszene <Programmierung>SprachsyntheseErhaltungssatzLie-GruppeWasserdampftafelDatensatzFrequenzURLEndliche ModelltheorieComputeranimation

10:45

Offene MengeProzess <Informatik>SpezialrechnerTotal <Mathematik>ServerKraftGruppenkeimKonfiguration <Informatik>BimodulSoftwaretestStichprobeGruppoidParametersystemLeistungsbewertungMigration <Informatik>Elektronische PublikationSchätzungWeb SitePaarvergleichRechenwerkAnalysisAttributierte GrammatikCodePunktPolygonTypentheorieFehlermeldungGesetz <Physik>KonditionszahlMultiplikationssatzLeistung <Physik>Objekt <Kategorie>Shape <Informatik>Ganze FunktionFrequenzÄhnlichkeitsgeometrieExplosion <Stochastik>UmwandlungsenthalpieInformationstechnikKanalkapazitätMobiles InternetSoftwareKonfiguration <Informatik>Prozess <Informatik>Bildgebendes VerfahrenOpen SourceFehlermeldungValiditätOrdnung <Mathematik>PasswortDatenbankDateiformatBitResultanteFunktionalLeistungsbewertungPaarvergleichPlug inStapeldateiParametersystemAbstandLokales MinimumQuellcodeGamecontrollerSoftwaretestOffene MengeRandwertDatenfeldMAPBenutzerbeteiligungEinfach zusammenhängender RaumClientEndliche ModelltheorieAdditionVerkehrsinformationMultifunktionAlgorithmische ProgrammierspracheGrenzschichtablösungMapping <Computergraphik>KreisflächeElektronische PublikationGeradeCASE <Informatik>ZeitrichtungCoxeter-GruppeOffice-PaketDesign by ContractMultiplikationsoperatorTaskInformationSpezielle unitäre GruppeMathematikFormation <Mathematik>SoftwareentwicklerDemoszene <Programmierung>Projektive EbeneBetrag <Mathematik>Migration <Informatik>Quick-SortGebäude <Mathematik>HypermediaElementargeometrieFlächeninhaltPhysikalisches SystemWald <Graphentheorie>Arithmetisches MittelComputeranimation

20:41

KanalkapazitätMobiles InternetInformationstechnikMultiplikationsoperatorMobiles InternetPhysikalisches SystemWeb ServicesBildgebendes VerfahrenWeg <Topologie>Ein-AusgabeOrtsoperatorInverser LimesFormale SpracheExpertensystemMengeSichtenkonzeptWellenpaketRechter WinkelDigitale PhotographieSmartphoneHumanoider RoboterInformationWeb-SeiteGebäude <Mathematik>CASE <Informatik>RichtungSchätzfunktionSoftwaretestSpeicherabzugProgrammierungZahlenbereichSpielkonsoleAnalysisFormation <Mathematik>AggregatzustandWorkstation <Musikinstrument>BildverstehenProgramm/QuellcodeComputeranimationBesprechung/Interview

Transkript: Englisch(automatisch erzeugt)

00:07

Okay, thank you. So I was told that I have a little bit more time because after that, this is lunch time, but I have no intention to take the lunch time, okay. So first of all, I'd like to show my teachers,

00:21

which was given four or five years ago at the Post 4G in Seoul. So this is the same color, but it's a different event. All right. Okay, and for this time, I'd like to give the presentation on the SPIRE 4, that means the Working Group 4 of UN Open GIS Initiative.

00:42

So this is for the Geospatial Data Collection. So there are many courses, most of them are from South Korea, from my university. And I'd like to also introduce Ananda Verkanan, who is working at the MINUSCA, MINUSCA, what is 10 for MINUSCA, exactly?

01:02

Yeah, so, but, so, so, Anna. So multi, so every time I hear that, every time I hear that, I forget it.

01:21

Sorry. But it is, so, it is based on Central Africa, Republic of Central Africa. So where there are many missions over there. And some of the courses like the GIT, GIT is the geospatial company in Korea who has been working for the validation

01:41

of a national mapping agency. And then Professor Kwang-Won Nam, who is, where is? Okay, so he is one of the leader of the product for SPIRE 4. I will give an explanation. And in Pyeong-Li, he is working at University of Seoul for the solution called LiveDromeMap.

02:03

All right, these are the also of this presentation, and we are working together. So I'd like to explain what is the goal of SPIRE 4 or Working Group 4. Working, so this is a life cycle of a geospatial system starting from the data construction.

02:23

So data construction is very beginning work and most fundamental things of geospatial system. Of course, after that, we have the management and the sharing of data and application and analysis. So everything start from this geospatial data collection.

02:42

So the activity of SPIRE 4 focused on the data collection. Of course, as you know, if you look at the spatial data structure of a single country or United Nations, there is not one single way

03:02

to collect the data. There are many, many ways. If you look at the mission field in Africa, so sometimes the situation of the data, the availability of geospatial data is very bad. So we have to struggle to find out the proper data,

03:22

any kind of geospatial data. So there is not only one single solution, but there are many, many solutions, like for example, migration of legacy data, if there is from proprietary software to open source solution or the, of course, the open stream map is one of the big source

03:42

of geospatial data for the United Nations. And so these are some activity within the SPIRE 4. But I would like to focus on the next three activity. First is a live-draw map and the second one

04:01

is geo-referenced multimedia data, called the GeoCMS, and the final one is geospatial data quality, so called OpenGeoDT. And of course, we include a vector tile map service, but which is led by Hidenori, so over there.

04:21

But he will give another presentation this afternoon about that activity. All right, so I will give the presentation of these three activity. So first one is GeoCMS. GeoCMS is led by Professor Nam over there. So the goal of this GeoCMS is to provide

04:41

geo-referenced multimedia, that means geo-referenced photo and geo-referenced video. So for example, we install a dashboard camera on the convoy vehicle, or the utility vehicle of the United Nations, then they capture every video, every frame of the video files

05:01

when they move to the mission field. So then we need a kind of a geo-reference, that means the trajectory of the video file to analyze and to find some information and so on. So the goal are to provide the solution for collecting and managing the geo-referenced multimedia.

05:24

Right, so in order to keep the geo-referenced data of a photo or the video, what is very important is to keep the information about the field of view. So as you know, if we take a picture,

05:41

we have a location, and we have the bearing, we have the heading, and the depths, and the angle, vertical angle, and horizontal angle. So we have to keep the metadata or the information about this FOV for every picture or every single frame of the video file. So that is something that we have to manage

06:02

in the system of Open GeoCMS. So we have two version of GeoCMS. The first one is just managing the data from the field of geo-referenced video and geo-referenced photo. So this is a case of geo-referenced photo,

06:21

and also if you look at the geo-referenced video, we have not only one single FOV, but the sequence of FOV, which correspond to the trajectory where you are moving with the camera. So we have to handle this kind of information. So this is the first version of GeoCMS.

06:42

Then after developing GeoCMS 1, of course we deliver it to the mission field of MINUSCA, and then we have been working for GeoCMS version 2, which is with some new feature, like a little bit modified user interface,

07:01

but most importantly, the image detection mechanism, function, for example, when you take a picture, then we can detect the person and the vehicle on the picture. So this is additional new function of GeoCMS version 2, and of course we implemented a little bit more complicated, sophisticated predicate,

07:23

so-called geospatial-temporal predicate. All right, so this is a comparison between version 1 and version 2. So we introduced, for example, the deep learning engine for the detection of the object using TensorFlow,

07:42

or M geometry library, M means mobile, moving object library. So there are many libraries for handling a spatial-temporal or moving object. So we included these two things into the new version of GeoCMS.

08:01

Okay, so this is a sample image of the feature detection using deep learning. So as you see here, it is not very clear from you, but we detected every persons here, and the vehicle, if there is. So this is the feature detection function,

08:21

and also we can give the query based on spatial-temporal condition. All right, and as a result of this system, we return the data with the JSON file. All right, so you can test this open GeoCMS version 2

08:40

via this website, all right. And the second one is a live-draw map. The main contributor is the team of Professor Lee. There are many Lees in Korea. Anyway, so Lee is one of the biggest family in the world since, okay, since nothing to do with my presentation.

09:03

All right, so Professor Lee in Pyeong, he is working at University of Seoul, and they are working together with Gaia 3D. The goal, and of course, Minuskai is, they are working for testing this solution. The goal is to deliver a solution

09:23

for collecting geo-rectified image captured from drone, and sharing via web portal. So there are two approaches. The first phase from 2016 and 2017, we focused on the option A,

09:40

which is to collect the data almost in real time from drone, and we process it almost in real time to produce a geo-rectified image map, and stitch it together, and provide it immediately via web portal. So one very good, very nice thing in option A is,

10:03

of course, the quality is not that good, but we can see what's happening on the image map, not only in the mission field, but, for example, in the UN headquarter, almost in real time, for example, every five seconds. So this is very good thing, but the quality is not that good,

10:22

so we have an option B, which is post-processing, because there are many very nice solution like OpenDRAW map, so we use OpenDRAW map to produce better quality image map, and to share via web portal through Mago3D.

10:43

Mago3D is a very nice engine to share dynamic data, so this is two approaches, so I will give a very short explanation of each. So first one is, okay, this is the requirement of the live drone map solution, so there are many requirement from the United Nation,

11:01

but anyway, so. This explain the whole process of the first option, the live drone map, so once you have, you prepare the region where you want to capture the data, then you launch the live drone map, then they send the image from each drone,

11:25

then it is processed in a ground, oh sorry. And then they generate geo-rectified data, and it is sent to geo-portal called Mago3D,

11:42

then this Mago3D data is shared by everywhere in the world via only a web browser, so this is very nice solution, but the problem was the quality was not that good. Anyway, we can see what's going on in the field. And then, so this is the comparison,

12:02

so this is some test using this solution, but I will skip it, yep. All right. So and for option two, because there is already very nice open source software called Open Drone Map,

12:21

so we use this one for testing five different region provided by Minuska, and then we tested it using web open drone map solution, probably some of you are already aware of that, so this is the result of this product.

12:43

If you see, this is not very clear, but you see the boundary of the image that we captured from the drone, and the other area, so this is very accurate, and we can get not only the geo-rectified image, but also the 3D modeling.

13:01

Of course, the 3D modeling is not very perfect, but anyway, we can get the idea of the terrain or the building and so on, so this is a very nice solution, and then after produced this solution that can be shared by every client in the world of the United Nations, almost seen.

13:23

All right, okay. So we installed this solution within Minuska, but we have some connection problem, but anyway, we are trying to install this one within Minuska. Okay, the next one is OpenGeoDT, which is a quality control solution for geospatial data.

13:47

This company, G-I-T, they have been working to validate the map produced by National Mapping Agency of Korea, so they have a lot of experience to validate the data.

14:01

So they applied this technique, and they released the source code as an open source of software, so the goal is to provide a quality control like validation, and if you find some errors, then you can correct it immediately,

14:20

and you provide as a result of the validation evaluation report. So this is the goal of this activity and web-based solution, but what they have done is they implemented this solution as a plugin of QGIS,

14:41

so this is quite a nice thing. And in order to run this quality control tool or validation tool, you have to set up some parameters. For example, if you want to detect the sliver line, you have to set the maximum distance

15:04

or minimum distance between two lines, et cetera. So you can set the parameters through the website, so these are the password anyway. So I'm sorry that document is available in Korean. So if you learn Korean, you can see that,

15:21

but anyway, this is very understandable without any knowledge on Korean, but anyway, we have to translate to English, but okay. So and this is explanation of the solution as a QGIS plugin. This is the comparison.

15:42

And one very interesting thing with GeoDTS, the solution OpenGeoDTS, we provided the quality control, not only for general data, but also OpenStreamMap data. All right, okay. So we first of all compared the validation items of JOSM.

16:01

JOSM is very nice tool to editing OpenStreamMap data. So we compared the functions or validation item from JOSM and the solution from GeoDTS. So of course there are common functions, but anyway, there are many, many very nice solutions, many items for the validation.

16:21

And we tested to compare our solution with JOSM. So we downloaded the data of South Sudan, OpenStreamMap of South Sudan, and we tested. So these are the results of the comparison. So of course the testing item of JOSM

16:41

and our solution are a little bit different, because the function for the validation are based on the JOSM format. But the GeoDT, we convert the OpenStreamMap data into the post-GIS, so these are based on the database on post-GIS.

17:01

The validation item are slightly different, but there are almost some correspondence between this function and this function. And this is a result of the test, so this is original file of OpenStreamMap, and the circle here, it indicates the region of the errors.

17:22

All right, these are the comparison of the result from JOSM and the OpenGeoDT. So there are some, almost these are very similar, except for example the tagging system, which are only in the OpenStreamMap data, and we don't have any tagging system in post-GIS.

17:43

So the checking, the tagging in post-GIS is meaningless, so we don't have that item. But anyway, these are almost the same. As a summary for the OpenGeoDT, the contribution is, so if you look at the process of JOSM, it is quite interactive checking.

18:02

It is not a batch processing. But for example, suppose that you are a national mapping agency, and you are subcontractor with the data collection company, then once they provide the data to a nation mapping agency, then we have to check

18:20

whether it satisfies the evaluation criteria. So these are such kind of purpose. So once we have OpenStreamMap data, probably Diego explained it in the previous presentation, they have to migrate this data periodically to UN-based map data.

18:42

So during this data, the migration, we need validate whether there are some errors in OpenStreamMap data or not. So these are for this process. All right, the summary of this presentation. Achievement, we developed several solutions

19:02

for geospatial data collection, and we delivered, and we are delivering as well. The things that we have to do is we have to, not install, we have to install within UN ICT infrastructure, because in order to install every software

19:21

within UN infrastructure, we have to follow a very rigorous procedure. So we need to follow that procedure. And the data collection process should be coupled, very tightly coupled with the workflow of United Nations. So they have their own procedure,

19:42

so that should be integrated or harmonized with the UN workflow. And we have to strengthen the community. So these solutions are very first step of the development, so we have to work together with the community from OSGEO and so on.

20:02

So for example, we are planning to set up an incubation project, not all, but some, via OSGEO. And what we need as well is we don't have yet the validation for field. That means suppose that you are the field officer,

20:20

then you have some work to do, but in addition to your own mission, you may carry your mobile solution, and you may check whether the data on the server or in the base map is correct or not. So if it is not correct, then you can very quickly make a report so that we may complete the data.

20:41

All right, that's it. Thank you. Thank you very much. We are again very well in time. Time for questions. Please, where's the first?

21:01

So the people? The link, the URL. Which URL, this one? This one. All right.

21:22

Okay, yeah, so to take picture. So there is the leader of the developing team, so Professor Nam, so if you have a question, then the technical question. So please do not ask me the technical question about that solution. He is the expert. What's next? All right, anyway, okay.

21:41

Regarding this tool, is it for only, for which kind of mobile system is, and in which language is it developed? So there is no limitation of the mobile devices. So once you get the FOB data, the position and the heading,

22:01

then you can get the iOS. So Professor Nam. So it is for iOS and Android? Some pages are available, but the system is not for the mobile system, just.

22:22

That is for the only web service in this time. Ah, okay, so web service. Oh, thank you, yes. This is, I think every smartphone is, okay, okay. Modern smartphone have magnetic sensors and GPS,

22:42

so you can take pictures that, that pictures have direction, including GPS, okay? So that is, we are using that information. So in the, when you take a video, okay, we, there is no magnetic sensor.

23:00

So we are casing that information from the direction, moving direction, okay? That is, we are, we can extract the information from the GPS information, okay?

23:27

Yes, yes, yes, okay. We are recognize where it is. Ah, yes, yes, it's possible. We, you can extract that information from the JPG, or all of the pictures, that is, okay. But that is available in the smartphone pictures,

23:44

smartphone pictures, and Android or iOS, okay, this is okay. Okay. Okay. I hope this, the next time. Yeah, hi, just one question. Are you doing any automatic feature extraction from the drone data? Drone data. Like 3D buildings, building outlines, roads, tracks.

24:02

Yeah, so that is very good question. So could you answer to this? So the automatic detection, not only from smartphone photo, but also the drone photo. Okay, ah, that is a running problem, so we have just two data set.

24:20

One is MS COCO, can you show that? This one? It's the MS COCO, MS COCO. MS COCO is for the, just the, our front, generally missing. The front view? Running set, the front view. That is not for the drone.

24:41

Oh, right. So, but we can make a new data set by the drones, and we can, we can running our system, and we can attach this system in our GeoCMS that is available, but not this time. Okay, so, so, so far you trained the system with the front image, yeah, but if you have enough

25:04

running training data set, then that could be also possible to realize for the drone image. Am I right? Our system is a cartridgeable system,

25:20

so we can add a MS COCO running data set for the front image, if you want to do data set, do running, do detection, okay, so you can, you can make a new running model, and add the system. That is available, but we have, we have very flexible system, but in this time,

25:40

that is just to support the MS COCO, MS COS, okay? So, so, okay, and. Okay, there was another one here, maybe? Ah, okay. Okay, so you are sold. Other questions? So, I do like to say that for data collection,

26:02

we also need your contribution. So this is very. That's clear, and it's usually nice to have a collective effort for that. If there are no more questions, I will stop the session here.