使用Kairos API进行人脸检测时遭遇参数无效错误求助

嘿，我一眼就看出问题所在啦——你收到的gallery_name is not a valid parameter和subject_id is not a valid parameter错误，是因为你把人脸注册接口的参数传给了人脸检测接口！

问题根源 #

Kairos的/detect接口是专门用来做人脸检测的，它只需要接收图片数据，不需要subject_id和gallery_name这两个参数；而这两个参数是给/enroll（人脸注册）接口用的，用来把人脸绑定到指定的用户ID和图库中。

两种修正方案，根据你的需求选择： #

方案1：如果你确实需要做人脸检测

直接去掉多余的values_enrol参数，只传递图片和认证头即可：

#!/usr/bin/python3
import requests
import glob, os
from os.path import basename

dir_path = "/home/Photos/"
path = "/home/Kairos/kairos_jason/"

os.chdir(dir_path)
for file in glob.glob("*.jpg"):
    img_path = file
    print(img_path)
    files = {'image': (img_path, open(dir_path + img_path, 'rb'), 'image/jpg', {'Expires': '0'})}
    headers = {
        'app_id': "My_Id",
        'app_key': "My_Key",
    }
    # 检测接口不需要subject_id和gallery_name，直接发送请求
    res = requests.post('https://api.kairos.com/detect', headers=headers, files=files)
    print("================RESPONSE=====================")
    print(res.content)
    strRes = str(res.content)
    print(strRes.find("face_id"))
    with open(path + os.path.splitext(img_path)[0] + ".txt" , 'w') as outfile:
        outfile.write(str(res.content))

方案2：如果你实际需求是注册人脸到图库

把请求的URL改成https://api.kairos.com/enroll，保留subject_id和gallery_name参数：

#!/usr/bin/python3
import requests
import glob, os
from os.path import basename

dir_path = "/home/Photos/"
path = "/home/Kairos/kairos_jason/"

os.chdir(dir_path)
for file in glob.glob("*.jpg"):
    img_path = file
    print(img_path)
    files = {'image': (img_path, open(dir_path + img_path, 'rb'), 'image/jpg', {'Expires': '0'})}
    headers = {
        'app_id': "My_Id",
        'app_key': "My_Key",
    }
    values_enrol = {
        "subject_id" : "11",
        "gallery_name" : "msc_galary"
    }
    # 注册接口使用enroll地址，支持subject_id和gallery_name参数
    res = requests.post('https://api.kairos.com/enroll', headers=headers, files=files, data=values_enrol)
    print("================RESPONSE=====================")
    print(res.content)
    strRes = str(res.content)
    print(strRes.find("face_id"))
    with open(path + os.path.splitext(img_path)[0] + ".txt" , 'w') as outfile:
        outfile.write(str(res.content))

额外说明 #

• /detect接口会返回人脸的坐标、年龄、性别等检测信息
• /enroll接口会将人脸特征存入指定图库，后续可以用/recognize接口进行人脸比对识别

内容的提问来源于stack exchange，提问作者Jayesh