* add implementation of RetinaNet

* * add README.md and model zoo
* rename FOCSFeat -> RetianFeat
* add mstrain model to model zoo
* refactor DeltaBBoxCoder

* update link for model and log

* fix doc link

* update doc, test=document_fix

* fix fuse_normalize not working when export model

* fix solov2 by keeping inputs setting before building model

* modify the ch_in&ch_out of the BasicBlock

(cherry picked from commit da568e4741f9a3a440ba1f605b8d12299fe4ed23)

* add the type convertion of the ch_in and ch_out & update gitignore

(cherry picked from commit 5b55097c6dab81dbccc187ba13a92b724ffc7ef5)

* revert the changes of gitignore

(cherry picked from commit ffeb72a8cefca6f416c5483756cc67f26f40522f)

This reverts commit 3e0e3585.

* fix picodet cpp infer

* Add enable_mkldnn to infer.py

* support export post_process in PicoDet

* add keypoint benchmark data

* update benchmark-data and change default keypoint bs=8, to be same as C++ infer

* support different norm_type for different module when convert to sync batch norm

* remove BatchNorm in esnet

* refine sync bn

* fix bugs of batch_norm

* fix bugs while deploying and modify BatchNorm to BatchNorm2D

* param_attr -> weight_attr in BatchNorm2D

* modify BatchNorm to BatchNorm2D

* fix checkpoint log when weight not loaded

* fix loading pretrain weight in mask rcnn

* discard unmatch id instead of remove

* support rcnn bs=2 in cpp infer

* check dynamic shape in padbatch

* replace fluid by paddle

* fix time display in pptracking cpp

* fix timer in bs=2

* add function wider face to coco dataset in x2coco.py