GitLab

Replace in-memory DastaSet with FixedLengthRecordDataSet

In-memory DataSet initiated without feed_dict are saved in TF graph def,
causing exported graph.pbtxt over-bloated.
Change to FixedLengthRecordDataSet and does a few refactorization.

* We do this because we want rules outside of slim to be able to depend on it in order to pull in slim models.

Fixing small typo in CONTRIBUTING.md

Add slim/scripts/finetune_inception_resnet_v2_on_flowers.sh

Make object_detection dictionaries compatible with Python3

Also fixes some typos in other scripts.

Update directories in running_pets.md

Correct documentation so all commands should be run from the root
of the git directory.

Update directories in running_pets.md

Also clarifies usage on label maps. In a nutshell, label map IDs
should start at index 1 (and not include 0).

Fix reversions plus a few improvements

Implementation of PCL and other newly proposed algorithms

* Change dictionnaries iteritems() to items()

CIFAR-10 Estimator-based example for multi-device training

Fix compatibility for object_detection/core/batcher.py

Fix CDNA transformation bug and speed up its implementation.

_strict_random_crop_image: fix mask slice

Minor typo obj detection's preparing_inputs.md

Make object detection's installation instruction more obvious

Also use the url link to github because *.md files are not
rendered when using jupyter notebook

- Fix CDNA transformation bug where transformed channels of color and masks were combined incorrectly.
- Remove for loop over batch size in implementation of CDNA transformation. This speeds up the building of the graph.

improved ArcStandardTransitionSystem.PerformRightArc()

Fixing evaluation config

Fix to [issue 1724](https://github.com/tensorflow/models/issues/1724)