Image Comprehension

import json
from tencentcloud.common import credential
from tencentcloud.common.profile.client_profile import ClientProfile
from tencentcloud.common.profile.http_profile import HttpProfile
from tencentcloud.mps.v20190612 import mps_client, models

cred = credential.Credential("<SecretId>", "<SecretKey>")
http_profile = HttpProfile()
http_profile.endpoint = "mps.tencentcloudapi.com"
client = mps_client.MpsClient(cred, "ap-guangzhou", ClientProfile(httpProfile=http_profile))

# 이미지 이해 태스크 제출 (모델 이름 지정 방식)
params = {
    "InputInfo": {
        "Type": "URL",
        "UrlInputInfo": {"Url": "https://example.com/document.jpg"}
    },
    "OutputStorage": {
        "Type": "COS",
        "CosOutputStorage": {"Bucket": "mybucket-125xxx", "Region": "ap-guangzhou"}
    },
    "OutputDir": "/output/comprehend/",
    "ScheduleId": 30200,
    "StdExtInfo": json.dumps({
        "ModelConfig": {
            "ModelName": "Google/gemini-2.5-flash",
            "Temperature": 0.8,
            "TopP": 0.9
        }
    }),
    "AddOnParameter": {
        "ExtPrompt": [{"Prompt": "Extract all text from the image"}]
    }
}

req = models.ProcessImageRequest()
req.from_json_string(json.dumps(params))
resp = client.ProcessImage(req)
print(resp.to_json_string())

# Definition 프리셋 방식: ScheduleId와 StdExtInfo 대신 Definition 사용
params.pop("ScheduleId")
params.pop("StdExtInfo")
params["Definition"] = 10004

# 태스크 결과 조회
query = models.DescribeImageTaskDetailRequest()
query.from_json_string(json.dumps({"TaskId": "<TaskId>"}))
detail = client.DescribeImageTaskDetail(query)
print(detail.to_json_string())

{
  "Response": {
    "TaskId": "2600007696-WorkflowTask-cDEF3456GH7890",
    "RequestId": "3c140219-cfe9-470e-b241-907877d6fb03"
  }
}

{
  "TaskId": "2600007696-WorkflowTask-cDEF3456GH7890",
  "Status": "FINISH",
  "CreateTime": "2025-05-21T10:00:00Z",
  "FinishTime": "2025-05-21T10:00:12Z",
  "Content": "This image shows an orange cat lying on a sunny windowsill, with a green garden in the background. The cat's eyes are half closed, and it looks very relaxed and comfortable."
}