[SoC] Just in time: Seeking mentors for development of a Deep Learning model applied to Remote Sensing Data

[Evandro Carrijo]

Hi Akbar,

The idea of implementing the RNN inside a Python plugin for QGIS is great.
Beside my experience as a QGIS user, there's already a QGIS plugin
<https://plugins.qgis.org/plugins/LAPIGTools/> developed by the lab where I
work, so I can take advantage of their expertise. I'll seriously consider
this possibility.

About the use of the temporal series, the information held in time between
observations has a great value for some targets. For example, it is quite
difficult to discriminate natural grasslands and pasturelands when relying
only on spatial information, but, in the other hand, the temporal signature
of the pixels holds precious information to classify between those two
classes. Other example is the identification of different land uses such
sugar cane and crops, which present big spectral variety in time and can
easily mislead a spatial model. All those targets are of great interest for
research in Brazil, and I am working on models trained on that kind of
classes.

Cheers, and thank you for the idea.

2018-03-22 17:50 GMT-03:00 Akbar Gumbira <akbargumbira at gmail.com>:

> Hi,
>
> It sounds like a nice idea. IMO you can also implement it as a python
> plugin for QGIS. If you need a simple RNN implementation, you can use
> Pyrenn (http://pyrenn.readthedocs.io/en/latest/) and ship it with the
> plugin.
>
> Just wondering, why does it need to consider the temporal information?
> What's the relevance of the history of a spatial area to the
> classification? Shouldn't it just classify based on the latest data?
>
> Cheers
>
> On Thu, Mar 22, 2018 at 5:49 PM, Evandro Carrijo <
> evandro.taquary at gmail.com> wrote:
>
>> Hello there!
>>
>> I'm a Computer Science Master's Degree student whose research if focused
>> on Deep Learning algorithms applied to Remote Sensing. Currently working at
>> the Laboratory of Image Processing and Geoprocessing
>> <https://github.com/lapig-ufg> settled at Federal University of Goiás -
>> Brazil. I'm also member of the High Performance Computing group of the same
>> university (more information here
>> <http://dgp.cnpq.br/dgp/espelhogrupo/7985061476854055>).
>>
>> Below I present an idea to explain how I can contribute to OSGeo
>> community and I'm seeking for mentors interested in assist my development.
>> Please, feel free to argue me any matter about the project idea.
>>
>> I would also appreciate a lot if you guys indicate a potential interested
>> mentor to my project idea or a OSGeo Project suitable to it.
>>
>> Hope there's some Interested ones out there!
>>
>> Idea
>>
>> The increasing number of sensors orbiting the earth is systematically
>> producing larger volumes of data, with better spatiotemporal resolutions.
>> To deal with that, better accurate machine learning approaches, such as
>> Deep Learning (DL), are needed to transform raw data into applicable
>> Information. Several DL architectures (e.g. CNN, semantic segmentation)
>> rely only at spatial dimension to perform, for example, land-cover/land-use
>> (LCLU) maps, disregarding the temporal dependencies between pixels
>> observations over the time. Also, high-res remote sensing data (e.g.
>> Planet, Sentinel) may provide more consistent time-series, that can be use
>> in the identification of important LCLU classes, like crop, pastureland and
>> grasslands.
>>
>> This potential can be explored using Recurrent Neural Networks (RNN), a
>> specific family of DL approaches which can take into account time
>> dimension. A promising project idea would be implement a RNN approach (e.g.
>> LSTM) to classify a Sentinel time-series, that will organize and preprocess
>> an input data set (e.g. labeled time-series), calibrate and evaluate a RNN
>> model, and finally classify an entire region (i.e. 2 or 3 scenes) to
>> produce a map for one or more LCLU class. It will be great evaluate the
>> accuracy and the spatial consistent of a map produced with a RNN approach.
>>
>> A simple example on classifying LCLU with two classes (pastureland and
>> non-pastureland):
>>
>> [image: itapirapua]
>> <https://user-images.githubusercontent.com/37085598/37687055-cc5236a8-2c78-11e8-8892-d113df44e235.jpg>
>> *Target region (input)*
>>
>> [image: itapirapua_ref]
>> <https://user-images.githubusercontent.com/37085598/37732806-ec792782-2d24-11e8-8ad9-18867768e998.jpg>
>> *Generated LCLU map (output)*
>> Best,
>>
>> Evandro Carrijo Taquary
>> Federal University of Goiás
>>
>> _______________________________________________
>> SoC mailing list
>> SoC at lists.osgeo.org
>> https://lists.osgeo.org/mailman/listinfo/soc
>>
>
>
>
> --
>
> *Akbar Gumbira *
> *www.akbargumbira.com <http://www.akbargumbira.com>*
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/soc/attachments/20180322/c224e0f9/attachment-0001.html>