* keypoint petr (#7774)

* petr train ok

train ok

refix augsize

affine size fix

update msdeformable

fix flip/affine

fix clip

add resize area

add distortion

debug mode

fix pos_inds

update edge joints

update word mistake

* delete extra codes;adapt transformer modify;update code format

* reverse old transformer modify

* integrate datasets

* add config and architecture for human36m (#7802)

* add config and architecture for human36m

* modify TinyPose3DHRNet to support human3.6M dataset

* delete useless class

---------
Co-authored-by: XYZ <1290573099@qq.com>

* [smdet] add tal_cr and custompan with transfomer, add box_distribution to pred reg_max, test=document_fix

* add tal_cr

* add ppyoloe_r

* modify code of ops.py

* add ppyoloe_r docs and modify rotate docs

* modify docs and refine connfigs

* fix some problems

* refine docs, add nms_rotated ext_op and fix some problems

* add image and inference_benchmark.py

* modify docs

* fix some problems

* modify code accroding to review

Co-authored-by: wangxinxin08 <>

* add fcosr

* fix some problem

* add docs for fcosr

* modify code

* modify focsr reader

* finish tensorrt deployment with dynamic shape

* modify according to review comment

Co-authored-by: wangxinxin08 <>

* 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

* update PicoDet Architecture and config

* [dev] add TOOD

* fix link and notes

* replace ConvNormLayer

* jde fairmot export and video python infer

* jde and deepsort: model

* 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.

* 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

* Rename object_detection to PaddleDetection
* Small fix for doc