‘face_alignment‘ has no attribute ‘FaceAlignment‘

最近需要提取人脸landmark,使用的是face-alignment:https://github.com/1adrianb/face-alignment

发现按照作者给出的Binaries 

安装方式总会报错

'face_alignment' has no attribute 'FaceAlignment'

因此切换为从source安装。

如果使用的是服务器,且没有root权限,那么最好为创建一个face-alignment的环境再安装。

# 环境搭建
git clone https://github.com/1adrianb/face-alignment
cd face_alignment
conda create -n FaceAlign python=3.7 # 这里版本任意3.5+即可
conda activate FaceAlign
pip install torch==1.12.1+cu116 torchvision==0.13.1+cu116 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu116

# 以下包按需安装
conda install matplotlib
conda install -c conda-forge ffmpeg
conda install -c conda-forge opencv

# 安装face_alignment
pip install -r requirements.txt
conda install -c 1adrianb face_alignment 
python setup.py install

codes of get landmarks

import face_alignment

from skimage import io


fa = face_alignment.FaceAlignment(face_alignment.LandmarksType.TWO_D, flip_input=False)

input = io.imread('/home/wyw/tes_imgs/src_0.jpg')

preds = fa.get_landmarks(input)

print(preds)

运行后会先下载模型,然后输出landmarks

$ CUDA_VISIBLE_DEVICES=9 python3 face_alignment.py
Downloading: "https://www.adrianbulat.com/downloads/python-fan/s3fd-619a316812.pth" to /home/wyw/.cache/torch/hub/checkpoints/s3fd-619a316812.pth
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 85.7M/85.7M [1:24:15<00:00, 17.8kB/s]
Downloading: "https://www.adrianbulat.com/downloads/python-fan/2DFAN4-cd938726ad.zip" to /home/wyw/.cache/torch/hub/checkpoints/2DFAN4-cd938726ad.zip
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 91.9M/91.9M [1:19:52<00:00, 20.1kB/s]
[array([[ 59., 119.],
       [ 59., 135.],
       [ 64., 153.],
       [ 66., 169.],
       [ 75., 182.],
       [ 84., 194.],
       [ 98., 198.],
       [111., 203.],
       [131., 207.],
       [154., 203.],
       [167., 200.],
       [179., 194.],
       [188., 182.],
       [194., 167.],
       [197., 151.],
       [199., 133.],
       [201., 117.],
       [ 77.,  88.],
       [ 84.,  79.],
       [ 95.,  74.],
       [104.,  74.],
       [113.,  77.],
       [147.,  77.],
       [156.,  74.],
       [165.,  74.],
       [176.,  79.],
       [185.,  88.],
       [131.,  97.],
       [131., 108.],
       [134., 117.],
       [134., 126.],
       [120., 137.],
       [127., 137.],
       [134., 140.],
       [138., 137.],
       [143., 137.],
       [ 91., 101.],
       [ 98.,  97.],
       [104.,  97.],
       [113., 101.],
       [107., 104.],
       [ 98., 104.],
       [147., 101.],
       [156.,  97.],
       [163.,  99.],
       [170., 101.],
       [165., 104.],
       [154., 104.],
       [111., 164.],
       [118., 158.],
       [129., 151.],
       [134., 153.],
       [138., 151.],
       [147., 158.],
       [156., 164.],
       [149., 171.],
       [140., 173.],
       [134., 173.],
       [125., 173.],
       [120., 169.],
       [111., 164.],
       [127., 160.],
       [134., 160.],
       [140., 160.],
       [154., 164.],
       [140., 164.],
       [134., 164.],
       [127., 164.]], dtype=float32)]

你可能感兴趣的:(人脸识别,深度学习,神经网络)