* fix run_benchmark for small model accurate speed

* fix run_benchmark for other det models

* add ppyoloe architectures

* fix deploy ppyoloe arch

* add ppyoloe arch coments, test=document_fix

* add centertrack base codes

* fix deploy and docs

* fix tracker

* fix

* fix pre_img

* fix deploy

* fix

* remove duplicate backbone variant, test=document_fix

* fix slice_image deploy infer index

* add scritps for npu and xpu

* rm slim_config in config

* rm enable lit egine for npu and xpu

* add lite_engine_enabled

* 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 <>

* fix slice_infer

* fix doc, test=document_fix

* add slice infer for smalldet

* add slice infer for smalldet

* fix slice infer

* fix doc, test=document_fix

* fix eval and configs

* add slice dataset loader

* fix deploy

* fix docs, test=document_fix

* update docs, test=document_fix

* add det based action, still in working

* add det_based action

* add id-based class action into pipeline

* remove clsactioninferer

* return history result, optimize visualization for multiple actions

* use history result for det-based action model

* batchinfer for cls&det models

* fix video infer(#5107)

* fix video infer(others)

* save detection results to file using coco format

* update save docs

* change id to image path when save result of image directory prediction

* fix error info

* fix infer dataset download and label

* add doc, test=document_fix

* add nms trt support

* add check version code

* fix bugs

* support mask rcnn bs>2

* fix resize op in c++ deploy

* fix preprocess in mask rcnn

* fix cascade mask rcnn when box is empty

* update merge batch

* 1. add timer for kpt& action 2. trt support for stgcn

* add parameter description

* make warm_frame configurable

* add keypoint infer and visualize into Pipeline

* add independent action model inference

* add action inference into pipeline, still in working

* test different display frames and normalization methods

* use bbox and scale normalization

* Remove debug info and Optimize code structure

* remove useless visual param

* make action parameters configurable

* add mkldnn bfloat16 args

* add mkldnn_bfloat16 to static deploy

* update

* update

* add bytetrack yolov3 engine and deploy baseline

* fix bytetrack yolov3 configs

* fix cfgs tools

* fix tracker pred_dets

* fit pptracking jde sde infer

* rename cfgs, fix mot_sde_infer

* fix mot sde single class infer

* remove byte cfgs

* fix mtmct

* clean code

* refactor det deploy

* refactor keypoint deploy

* fix solov2

* fit mot sde pipeline infer

* refine mot sde infer

* fit mot jde infer pipeline

* fit mot pose unite infer

* precommit for format

* refine keypoint detector name

* clean codes

* fix keypoint infer

* refine format
Co-authored-by: NFeng Ni <nemonameless@qq.com>

* 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