GitLab

* Creates a new batch_decode method in SSD Meta architecture that can handle
  dynamic batch size.
* use combined_shapes in _get_feature_maps_spatial_dims method to handle
  dynamic batch image_size.
* Add dynamic batch size tests to check preprocess, predict and postprocess
  methods in SSD Meta architecture.

* Adds a util function to compute a mix of dynamic and static shapes
  preferring static when available.
* Uses batch_multiclass_non_max_suppression function in postprocess_rpn
  instead of looping over static batch shape and performing
  multiclass_non_max_suppression.
* Adds a new helper function _unpad_proposals_and_sample_boxclassifier_batch
  to sample from a batch of tensors possibly containing paddings.
* Tests batch inference with various configurations of static shape via
  unittests.

A few change to prepare for batch inference:

* Modify the return type of batch non max suppression to be tuple of tensors
  so it can be reused for both stages of faster rcnn without any confusion
  in the semantics implied the the keys used to represent the tensors.
* Allow dynamic number of anchors (boxes) in addition to dynamic batch size.
* Remove a redundant dynamic batch size test.

This change is required to enable object detection export and inference
with dynamic batch size.

Bug:
We use two different Saver objects during training, one to initialize from a
checkpoint and another to save and restore from checkpoints during training.
Both these Savers were bound to same variable name causing init_fn function
to use the latter instead of the former when called.

Adds an op to handle pre-computed word embeddings.

use external url for mlengine link

Fixes #1841

Fix indentation in obj detection tutorial

This change adds an op that will take a text vocabulary file, read
it, and use it to create a pre-computed word embedding vector.

Removed wrong parameter in another example

Fix import on export_inference_graph_test.

Update directories in preparing_inputs.md

Removed an error-causing parameter

modified the last line in download_and_convert_data.py scripts

Update Readme Install Instructions

Change split name in slim's export_inference_graph

Spatial attention for the Attention OCR model.

Update slim/nets.

removing memory intensive session.graph from FileWriter call

Add CODEOWNERS to auto-assign PRs

Extracted from first commit for each directory. Should be a reasonably good proxy.

Add back six dependency and change pbtxt files to reflect new way of …

- Set spatial squeeze to true for resnets
- Make activations configurable for resnets
- Make batch norm configurable for resnets
- Fix some bad comment wrapping in vgg.py.

Replace in-memory DastaSet with FixedLengthRecordDataSet