python - Azure 计算机视觉 SDK read_in_stream 始终返回错误请求

Question

我发现自己无法在 Python 中使用 Azure 计算机视觉 SDK 的 read_in_stream API。SDK 总是返回一个

File "<path>/azure/cognitiveservices/vision/computervision/operations/_computer_vision_client_operations.py", line 1567, in read_in_stream raise models.ComputerVisionErrorException(self._deserialize, response) azure.cognitiveservices.vision.computervision.models._models_py3.ComputerVisionErrorException: Operation returned an invalid status code 'Bad Request'

我尝试使用标准读取 API，为其提供文件 URL 而不是本地 URL，效果很好，因此客户端已得到正确身份验证。

该文件确实具有读取权限，并且将其读入枕头图像效果很好。这是一个相当小的 PNG 文件。

from azure.cognitiveservices.vision import computervision
from azure.cognitiveservices.vision.computervision import ComputerVisionClient
from azure.cognitiveservices.vision.computervision.models import OperationStatusCodes
from azure.cognitiveservices.vision.computervision.models import VisualFeatureTypes
from msrest.authentication import CognitiveServicesCredentials

from array import array
import os
from PIL import Image
import sys
import time

subscription_key = "<key from azure portal>"
endpoint = "<endpoint from azure portal>"


def ocr(filePath):

    computervision_client = ComputerVisionClient(endpoint, CognitiveServicesCredentials(subscription_key))

    with open(filePath, "rb") as local_image_printed_text:
        recognize_handw_results = computervision_client.read_in_stream(local_image_printed_text, raw=True)

Answer 1

您提到您的 PNG 文件很小，请您检查一下它的尺寸是否大于 5050？基于official API reference : 如果使用尺寸小于 5050 的 PNG 文件，我可以重现您的问题:

但是当我换成更大的时，它的工作方式异常(exception):

这是我的测试代码:

from azure.cognitiveservices.vision import computervision
from azure.cognitiveservices.vision.computervision import ComputerVisionClient
from azure.cognitiveservices.vision.computervision.models import OperationStatusCodes
from azure.cognitiveservices.vision.computervision.models import VisualFeatureTypes
from msrest.authentication import CognitiveServicesCredentials

from array import array
import os
from PIL import Image
import sys
import time
import requests

subscription_key = ""
endpoint = ""


def ocr(filePath):

    computervision_client = ComputerVisionClient(endpoint, CognitiveServicesCredentials(subscription_key))

    with open(filePath, "rb") as local_image_printed_text:
        recognize_handw_results = computervision_client.read_in_stream(local_image_printed_text, raw=True)
        
        time.sleep(5)
        result_url = recognize_handw_results.headers.get('Operation-Location')
        result = requests.get(result_url,headers = {"Ocp-Apim-Subscription-Key":subscription_key})
        print(result.text)

ocr("<png local path>")