实战OpenPose项目4：实时准确的全身多人姿态估计和跟踪系统
官网：https://github.com/MVIG-SJTU/AlphaPose
源码使用google服务器运行：https://colab.research.google.com/drive/14Zgotr2_F0LfvcpRi03uQdMvUbLQSgok?usp=sharing#scrollTo=3VBhQTOSoWab
推理过程：
! pip install pyyaml==5.2
! pip install scipy==1.1.0
! pip install torch==1.2.0 torchvision==0.4.0
! pip install pillow==6.2.2
import torch
print(torch.__version__)
import yaml, scipy
print(yaml.__version__)
print(scipy.__version__)
!rm -rf /content/AlphaPose
import os
os.chdir('/content/')
!git clone https://github.com/MVIG-SJTU/AlphaPose.git
!python -m pip install cython
!sudo apt-get install libyaml-dev
import os
os.chdir('/content/AlphaPose')
print(os.getcwd())
! python setup.py build develop
!pip install -U -q PyDrive
from pydrive.auth import GoogleAuth
from pydrive.drive import GoogleDrive
from google.colab import auth
from oauth2client.client import GoogleCredentials
# Authenticate and create the PyDrive client.
# This only needs to be done once per notebook.
auth.authenticate_user()
gauth = GoogleAuth()
gauth.credentials = GoogleCredentials.get_application_default()
drive = GoogleDrive(gauth)
! mkdir /content/AlphaPose/detector/yolo/data
file_id = '1D47msNOOiJKvPOXlnpyzdKA3k6E97NTC'
downloaded = drive.CreateFile({'id': file_id})
downloaded.GetContentFile('/content/AlphaPose/detector/yolo/data/yolov3-spp.weights')
! mkdir /content/AlphaPose/detector/tracker/data
file_id = '1nlnuYfGNuHWZztQHXwVZSL_FvfE551pA'
downloaded = drive.CreateFile({'id': file_id})
downloaded.GetContentFile('/content/AlphaPose/detector/tracker/data/JDE-1088x608-uncertainty')
file_id = '1kQhnMRURFiy7NsdS8EFL-8vtqEXOgECn'
downloaded = drive.CreateFile({'id': file_id})
downloaded.GetContentFile('/content/AlphaPose/pretrained_models/fast_res50_256x192.pth')
import os
os.chdir('/content/AlphaPose')
! ls
! python3 scripts/demo_inference.py --cfg configs/coco/resnet/256x192_res50_lr1e-3_1x.yaml --checkpoint pretrained_models/fast_res50_256x192.pth --indir examples/demo/ --save_img
# result json and rendered images are saved here:
! ls examples/res/
! ls examples/res/vis
官方步骤：
快速开始
-
Colab：我们提供了一个colab 示例供您快速入门。
-
推理：推理演示
./scripts/inference.sh ${CONFIG} ${CHECKPOINT} ${VIDEO_NAME} # ${OUTPUT_DIR}，可选
高级API请参考 ./scripts/demo_api.py
- 训练：从头开始训练
./scripts/train.sh ${CONFIG} ${EXP_ID}
- 验证：在 MSCOCO val2017 上验证您的模型
./scripts/validate.sh ${CONFIG} ${CHECKPOINT}
例子：
演示使用FastPose模型。
./scripts/inference.sh configs/coco/resnet/256x192_res50_lr1e-3_1x.yaml pretrained_models/fast_res50_256x192.pth ${VIDEO_NAME}
#或
python 脚本/demo_inference.py --cfg/configs/210r16e.yaml configs/21r3_16_16re -checkpoint pretrained_models/fast_res50_256x192.pth --indir examples/demo/
列车FastPose上mscoco数据集。
./scripts/train.sh ./configs/coco/resnet/256x192_res50_lr1e-3_1x.yaml exp_fastpose
更详细的推理选项和示例，请参考GETTING_STARTED.md