* add devices_num on reader

* refine capacity value

* refine bufsize

* add multi-scale testing function in cascade rcnn clsaware architecture

* Polish Reader to simplify preprocessing logic.
#  sample_transforms-> make batch -> batch_transforms in Reader.
* Clean some code
* Imporve yolov3_r50vd_dcn_obj365_pretrained_coco 41.4 to 41.8.
* Update all configs.

* add multi-scale test

* 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 3a2c106271885071db7c0d85587540a8f83c24db.

* 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

* minor fix for PaddleDetection

* fix rpn_only in cascade mask fpn

* fix rpn_only in cascade rcnn

* refine cascade models

* add rpn recall function and class-aware rpn

* Rename object_detection to PaddleDetection
* Small fix for doc