* add caption model ETS for PaddleVideo

* modify dynamic rnn to static rnn, and add infer past

* modify reader and configs for evaluation and validation

* modify details of the ets net code

* modify API of data and dataloader for ets net

* replace instance_norm

* add version check

* update check version

* -1->None

* fix mistake

* refine assert info in roidb_source.py

* refine image_dir assert

* fix fluid.data and fluid.io.DataLoader

* add logger in utils/utility.py

* Add face detection configs and main code.
* Integration multi-scale evaluation.

* unify reader to dataloader

* add check_version

* change PyReader to from_generator

* fluid.layers.data->fluid.data

* fix_bug

* add_multiprocess_train_to_image_classification

* Document mixed precision training flags

* Clarify GPU is required for mixed precision training

* refine build strategy for the model optimize test=develop

* fix according develop test=develop

* add CropImageWithDataAnchorSampling
* Refine bbox sampling code
* Update comments
* fix import and variable problem

* Add backbone and architecture for FaceBoxes and Blazeface.
* Add dataloader for WIDERFace dataset.

Update ppdet/utils/checkpoint.py to skip the var shape check when the var is not initialized. (#3452)

* Place mixed precision inside PaddleDetection

roll back to the monkey patch version as a temporary measure, before it is
merged into paddle

* Add command flag for `loss_scale`

* Fix a stupid indentation error

optimizer should be in the mixed precision context

* Initial FP16 training

* Add mixed precision training to rest of the detection models

* Revert "Add support for mixed precision training (#3406)"

This reverts commit 3a2c1062.

* Bug fixes and some tweaks

* Support finetune by custom dataset

* add finetune args

* add load finetune

* reconstruct load

* add transfer learning doc

* add fruit demo

* add quick start

* add data preprocessing FAQ