封装face_recognition制作成Docker，提供裁剪人像与人像比对相似度两个接口
# 构建镜像
sudo docker build -t fr-compare:2.0.0 .
# 启动容器
sudo docker run --restart=always --name fr-compare -d -p 8726:8726 fr-compare:2.0.0
访问http://0.0.0.0:8726, 页面输出
{"code":200,"message":"face_recognition server 2.0"}
http://0.0.0.0:8726/detect （POST）
请求头
'Content-Type': 'application/json'
请求体
{
"b64_image": "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQEC..."
}
回复内容
{
"b64_image": "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQEC...", "message": "success"
}
返回的 b64_image 为裁剪后的头像数据, message 为 "succcess" 时为截取头像成功, 否则为报错信息并且b64_image 为空字符串
http://0.0.0.0:8722/detect （POST）
请求头
'Content-Type': 'application/json'
请求体
{
"b64_image_01": "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQEC..."，
"b64_image_02": "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQEC..."
}
回复内容
{
"message": "ok",
"distance": 56.84
}
distance 值越高，两张照片为一个人的概率越大
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent