* given inference model path, download model automatically

* encapsulate auto download model as a function

* given note that default model dir is download link

* change attr model download path to https://bj.bcebos.com/v1/paddledet/models/pipeline/PPLCNet_x1_0_person_attribute_945_infer.zip

* Update action.md

delete the step of change model name of fight recognition

* Update video_action_infer.py

get model path from model_dir without model name

* optimize action vis

* Scale frames before fight action recognition

* put
short_size = self.cfg["VIDEO_ACTION"]["short_size"]
scale = Scale(short_size)
out of while

* change class name from Scale to ShortSizeScale

* add fight for PP-Human

* add short_size and target_size for fight recognition

* add short_size and target_size for fight_infer

* modify code according to the reviews

* add the wrong deleted lines`

* Update pipeline.py

* Update infer_cfg.yml

* visualize fight when fight action occur

* 乱码修改

* delete useless parmas

* delete useless code str2bool