This repository refactors the inference from the official implementation of DWPose.
# git clone this repository
git clone https://github.com/haofanwang/DWPose.git
cd DWPose
# install required packages
pip install -r requirements.txt
# Set environment
pip install -U openmim
mim install mmengine
mim install "mmcv>=2.0.1"
mim install "mmdet>=3.1.0"
mim install "mmpose>=1.1.0"If you meet any download issues, please refer to installation instructions.
Download the pretrained detection model and pose model, and save them under ./ckpts. It is also possible to use other detection models from MMDetection and pose model.
import cv2
import numpy as np
from PIL import Image
from utils import *
from dwpose import DWposeDetector
# set configs
det_config = './dwpose/yolox_config/yolox_l_8xb8-300e_coco.py'
det_ckpt = './ckpts/yolox_l_8x8_300e_coco_20211126_140236-d3bd2b23.pth'
pose_config = './dwpose/dwpose_config/dwpose-l_384x288.py'
pose_ckpt = './ckpts/dw-ll_ucoco_384.pth'
# set device
device = "cuda:0"
# init
dwprocessor = DWposeDetector(det_config, det_ckpt, pose_config, pose_ckpt, device)
# infer
image_dir = "./assets/test.jpeg"
input_image = cv2.imread(image_dir)
input_image = HWC3(input_image)
input_image = resize_image(input_image, resolution=512)
detected_map = dwprocessor(input_image)
detected_map = HWC3(detected_map)
detected_map = cv2.resize(detected_map, (W, H), interpolation=cv2.INTER_LINEAR)
cv2.imwrite(image_dir.split('/')[-1], detected_map)