github链接:https://github.com/JianqiangRen/FlowBasedBodyReshaping

一、介绍:

我们提出了一种用于人体重塑 (FBBR) 的端到端结构感知流生成框架,可以有效地为高分辨率图像取得有利且可控的结果。BR-5K 是第一个用于身体重塑的大规模数据集,它由从Unsplash收集的 5,000 张 2K 分辨率的高质量个人肖像照片组成

二、下载模型

文章百度链接提供

https://pan.baidu.com/share/init?surl=hxlh5jipyaEi1R6NLm8KFg&pwd=17e6

(key:17e6)

下载模型放入models目录

三、配置环境

conda create -n FlowBasedBodyReshaping python=3.8
conda activate FlowBasedBodyReshaping
pip install "torch-1.11.0+cu113-cp38-cp38-win_amd64.whl"
pip install "torchvision-0.12.0+cu113-cp38-cp38-win_amd64.whl"
pip install opencv-python -i https://mirror.baidu.com/pypi/simple/
pip install numba -i https://mirror.baidu.com/pypi/simple/
pip install easydict -i https://mirror.baidu.com/pypi/simple/
pip install scipy -i https://mirror.baidu.com/pypi/simple/
pip install matplotlib -i https://mirror.baidu.com/pypi/simple/
pip install tqdm -i https://mirror.baidu.com/pypi/simple/
pip install lpips -i https://mirror.baidu.com/pypi/simple/
pip install toml -i https://mirror.baidu.com/pypi/simple/

四、运行

1、配置文件解释,打开test_demo_setting.toml

degree = 3.0

suppress_bg = false

flow_scales = ['upper_2']

divider = 20

reshape_ckpt_path = 'models/body_reshape_model.pth'


src_dir = './test_cases'
gt_dir = 'null'
save_dir = 'output'

2、运行命令

python test.py --config config/test_demo_setting.toml

五、瘦身效果