Google Cloud Vision API 使用 Cloud Shell：如何运行 API 用于多个图像？我的 request.json 应该是什么样子？

Question

我运行对单个图像进行了测试 [使用 Cloud Shell]，request.json 如下所示。我怎样才能运行 Vision API 整个文件夹的图像？另外，为什么API到运行的图片用户权限需要是public？谢谢。

{
"requests": [
      {
"image": {
"source": {
"gcsImageUri": "gs://visionapitest/landmark/test.jpeg"
          } 
        },
"features": [
          {
"type": "LABEL_DETECTION",
"maxResults": 10
          }
        ]
      }
  ]
}

Answer 1

如果你想使用云端执行请求Shell，你必须按照以下方式进行

{
"requests": [
{
  "image": {
    "source": {
      "gcsImageUri": "gs://visionapitest/landmark/test.jpeg"
    }
  },
  "features": [
    {
      "type": "LABEL_DETECTION",
      "maxResults": 10
    }
  ]
},
{
  "image": {
    "source": {
      "gcsImageUri": "gs://visionapitest/landmark/test2.jpeg"
    }
  },
  "features": [
    {
      "type": "LABEL_DETECTION",
      "maxResults": 10
    }
  ]
}, … ]}

请注意，这不是指定完整文件夹的方法，因为您可以看到“请求”字段是 AnnotateImageRequest 个对象的数组，因此您必须逐项列出 JSON 文件.

另一方面，您可以使用一个可用的 Vision Client Libraries in order to read all the images within the folder. I would like to share a python code snippet I took from the Vision API documentation 动态创建“requests”数组，虽然它只考虑了一张图片，但我修改它以读取整个文件夹。

from google.cloud import vision_v1
from google.cloud.vision_v1 import enums
from google.cloud import storage
from google.cloud.vision_v1 import types
from re import search

def sample_async_batch_annotate_images(
   bucket_name,
   output_uri
):
"""Perform async batch image annotation."""
client = vision_v1.ImageAnnotatorClient()

storage_client = storage.Client()
blobs = storage_client.list_blobs(
    bucket_name, prefix='vision/label/', delimiter='/'
)

requests = []

for blob in blobs:
    if search('jpg',blob.name):
        input_image_uri = 'gs://' + bucket_name +'/'+ blob.name
        print(input_image_uri)

        source = {"image_uri": input_image_uri}
        image = {"source": source}
        features = [
            {"type": enums.Feature.Type.LABEL_DETECTION},
        ]

        request = types.AnnotateImageRequest(image=image, features=features)
        requests.append(request)

gcs_destination = {"uri": output_uri}

# The max number of responses to output in each JSON file
batch_size = 2
output_config = {"gcs_destination": gcs_destination,
                 "batch_size": batch_size}

operation = client.async_batch_annotate_images(requests, output_config)

print("Waiting for operation to complete...")
response = operation.result(90)

# The output is written to GCS with the provided output_uri as prefix
gcs_output_uri = response.output_config.gcs_destination.uri
print("Output written to GCS with prefix: {}".format(gcs_output_uri))

不过，您可以将此作为参考，但这取决于您的 use-case 和代码语言偏好。

关于权限的问题，我想你指的是云存储桶的。根据我的理解，没有必要制作您的图像 public，您只需将存储桶中的 read/write Cloud Storage permissions 提供给您正在执行请求的服务帐户。

Google Cloud Vision API 使用 Cloud Shell：如何运行 API 用于多个图像？我的 request.json 应该是什么样子？

Google Cloud Vision API using Cloud Shell: How can I run the API for multiple images? What should my request.json look like?

google-cloud-platform