如何访问Azure Image Analysis 4.0 API？遇404资源未找到错误求助

问题背景 #

部署在美国东部（USA EAST）区域的Azure Image Analysis 4.0资源，通过Python代码调用REST API时持续返回404 "Resource not found"错误。

实现代码 #

import os
import requests
import pyautogui
from PIL import Image
from io import BytesIO
from azure.cognitiveservices.vision.computervision import ComputerVisionClient
from azure.cognitiveservices.vision.computervision.models import VisualFeatureTypes
from msrest.authentication import CognitiveServicesCredentials

# Azure Configuration
SUBSCRIPTION_KEY = "..."
ENDPOINT = "..."

# OpenAI Configuration
OPENAI_KEY = "..."

def take_screenshot():
    screenshot = pyautogui.screenshot()
    output = BytesIO()
    screenshot.save(output, format="PNG")
    output.seek(0)  # Rewind the stream back to the beginning before reading from it
    return output

def analyze_image(image_stream):
    headers = {
        'Content-Type': 'application/octet-stream',
        'Ocp-Apim-Subscription-Key': SUBSCRIPTION_KEY,
    }
    params = {
        'visualFeatures': 'Objects,Tags,Description,Faces,ImageType,Color,Adult',
        'language': 'en',
    }
    response = requests.post(ENDPOINT + 'vision/v4.0/analyze', headers=headers, params=params, data=image_stream)
    print(f"Status code: {response.status_code}")
    print(f"Response text: {response.text}")
    response.raise_for_status()
    return response.json()

def query_openai(azure_result):
    headers = {
        'Authorization': f'Bearer {OPENAI_KEY}',
        'Content-Type': 'application/json',
    }
    
    # Format the Azure result into a string
    result_string = "Description: {}. Tags: {}. Objects: {}. Adult content: {}. Racy content: {}.".format(
        azure_result['description']['captions'][0]['text'] if azure_result['description']['captions'] else "None",
        ', '.join([tag['name'] for tag in azure_result['tags']]),
        ', '.join([obj['objectProperty'] for obj in azure_result['objects']]) if 'objects' in azure_result and azure_result['objects'] else "None",
        azure_result['adult']['isAdultContent'],
        azure_result['adult']['isRacyContent']
    )

    print(f"Azure details sent to OpenAI: {result_string}")  # Print the Azure details

    data = {
        'model': 'gpt-4',
        'messages': [
            {"role": "system", "content": "Translate the following Azure Image Analysis result to a boolean value."},
            {"role": "user", "content": f"{result_string}. Does the image contain an apple?"}
        ]
    }
    
    response = requests.post('https://api.openai.com/v1/chat/completions', headers=headers, json=data)
    return response.json()['choices'][0]['message']['content'].strip().lower() == 'true'

if __name__ == "__main__":
    image_data = take_screenshot()
    azure_result = analyze_image(image_data)
    is_apple_present = query_openai(azure_result)
    print('Does the screen have an apple on it?', is_apple_present)

错误响应 #

Status code: 404
Response text: {"error":{"code":"404","message": "Resource not found"}}
Traceback (most recent call last):
  File "C:\Users\...\Desktop\ScreenAnalysisApp\app.py", line 69, in <module>
    azure_result = analyze_image(image_data)
                   ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\...\Desktop\ScreenAnalysisApp\app.py", line 36, in analyze_image
    response.raise_for_status()
  File "C:\Users\...\AppData\Local\Programs\Python\Python311\Lib\site-packages\requests\models.py", line 1021, in raise_for_status
    raise HTTPError(http_error_msg, response=self)
requests.exceptions.HTTPError: 404 Client Error: Resource Not Found for url: https://purefocus.cognitiveservices.azure.com/vision/v4.0/analyze?visualFeatures=Objects%2CTags%2CDescription%2CFaces%2CImageType%2CColor%2CAdult&language=en

排查与解决 #

核心问题：API端点路径与参数不匹配 #

Azure Image Analysis 4.0使用了全新的API端点和参数格式，旧的/vision/v4.0/analyze路径属于Computer Vision v3.x版本，不适用于4.0版本。

具体修正步骤 #

1. 更换API端点路径
   将代码中的ENDPOINT + 'vision/v4.0/analyze'替换为ENDPOINT + 'computervision/imageanalysis:analyze'，并添加api-version参数（推荐使用2023-02-01-preview或更新的GA版本）。

2. 调整请求参数
   4.0版本用features替代了旧的visualFeatures参数，且feature名称为小写格式，比如caption代替Description、objects代替Objects。

3. 适配返回的JSON结构
   4.0版本的响应结构与旧版本不同，比如描述信息在captionResult字段下，标签在tagsResult.values下，需要同步调整后续处理逻辑。

修正后的关键代码片段 #

修正后的analyze_image函数

def analyze_image(image_stream):
    headers = {
        'Content-Type': 'application/octet-stream',
        'Ocp-Apim-Subscription-Key': SUBSCRIPTION_KEY,
    }
    params = {
        'features': 'objects,tags,caption,faces,imageType,color,adult',
        'language': 'en',
        'api-version': '2023-02-01-preview'
    }
    # 使用4.0版本的正确路径
    response = requests.post(ENDPOINT + 'computervision/imageanalysis:analyze', headers=headers, params=params, data=image_stream)
    print(f"Status code: {response.status_code}")
    print(f"Response text: {response.text}")
    response.raise_for_status()
    return response.json()

修正后的query_openai函数中的结果格式化部分

# 适配4.0版本的响应结构
result_string = "Description: {}. Tags: {}. Objects: {}. Adult content: {}. Racy content: {}.".format(
    azure_result['captionResult']['text'] if azure_result['captionResult']['text'] else "None",
    ', '.join([tag['name'] for tag in azure_result['tagsResult']['values']]),
    ', '.join([obj['name'] for obj in azure_result['objectsResult']['values']]) if 'objectsResult' in azure_result and azure_result['objectsResult']['values'] else "None",
    azure_result['adultResult']['isAdultContent'],
    azure_result['adultResult']['isRacyContent']
)

额外验证点 #

• 确认Azure资源是Computer Vision（单服务或多服务），且处于正常运行状态。
• 验证ENDPOINT配置正确，未遗漏末尾的斜杠（如果你的ENDPOINT没有斜杠，拼接时需要加上，比如ENDPOINT.rstrip('/') + '/computervision/...'）。

内容的提问来源于stack exchange，提问作者Troy Robinson