[MXNET-102] Added tutorial on how to use data augmenters.

[thomelane]

Description

Added tutorial on methods of using data augmenters.

Checklist

Essentials

N/A, as only added a single markdown file.

Changes

Added tutorial.

Comments

Quote from tutorial:

In this tutorial we discuss the different interfaces avaliable in MXNet to perform data augmentation. We start by showing a complete example using Gluon's mxnet.gluon.data.Dataset, and then unpack the example to gain a greater understanding of the internals. In the process you'll learn about augmentation functions, mxnet.image.Augmenter classes and Augmenter lists. We also provide examples using mxnet.image.ImageIter if you haven't yet moved to Gluon.

[rahul003]

The filename and location (docs/tutorials/python/using_data_augmenters/Methods of applying data augmentation.md) are awkward.

Could you make it simpler like docs/tutorials/python/data_augmentation.md ?

[cjolivier01]

Please add a JIRA

[thomelane]

@rahul003 made the suggested changes.
@cjolivier01 added JIRA task https://issues.apache.org/jira/browse/MXNET-102

[thomelane]

@rahul003 @szha @piiswrong @cjolivier01 can we get this merged?

[piiswrong]

Mixing gluon and mx.image is weird. We should have separate tutorials

[thomelane]

@piiswrong Until we've got the new Gluon augmentation, using mx.image inside a transform function is the best approach I though? It's the only way I've found in the examples.

[thomelane]

@marcoabreu the build seems temperamental at the moment, got a successful build on the 4th attempt. My 1st attempt errored trying to download CIFAR-100, 2nd attempt failed on test_gluon.test_lambda as Error 1.270788 exceeds tolerance rtol=0.001000, atol=0.000000. and the 3rd time I couldn't quite tell exactly what was the issue but build.py: 2018-04-05 17:39:40,153 Running of command in container failed (255): docker run --rm -v /home/jenkins_slave/workspace/ut-perl-cpu:/work/mxnet -v /home/jenkins_slave/workspace/ut-perl-cpu/build:/work/build -u 1001:1001 mxnet/build.ubuntu_cpu /work/runtime_functions.sh unittest_ubuntu_cpugpu_perl. Any idea what's causing this given my changes are only markdown?

[marcoabreu]

Yeah sorry, our tests are quite flaky :/

[thomelane]

@marcoabreu are there Github/JIRA issue(s) for the individual tests that fail? or is there a general issue I can add this information to?

[marcoabreu]

Yes, we track them on GitHub with the labels "Flaky" and "test". I think they're all tracked (pexcept the perl ones), but feel free to double check if the ones you encountered are already there.