dsProject/dsRagAnything/ShiTi/T1_QWenOcr.py

from openai import OpenAI

client = OpenAI(
    api_key='sk-f6da0c787eff4b0389e4ad03a35a911f',
    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
)

prompt = "请提取图片中的试题"
completion = client.chat.completions.create(
    model="qwen-vl-ocr-latest",
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "image_url",
                    "image_url": "https://pic1.zhimg.com/v2-c3d7f060bbf3f7c122319350044d8888_1440w.jpg",
                    # 输入图像的最小像素阈值，小于该值图像会按原比例放大，直到总像素大于min_pixels
                    "min_pixels": 28 * 28 * 4,
                    # 输入图像的最大像素阈值，超过该值图像会按原比例缩小，直到总像素低于max_pixels
                    "max_pixels": 28 * 28 * 8192
                },
                # qwen-vl-ocr-latest支持在以下text字段中传入Prompt，若未传入，则会使用默认的Prompt：Please output only the text content from the image without any additional descriptions or formatting.
                # 如调用qwen-vl-ocr-1028，模型会使用固定Prompt：Read all the text in the image.不支持用户在text中传入自定义Prompt
                {"type": "text",
                 "text": prompt},
            ]
        }
    ])

print(completion.choices[0].message.content)

# 将返回的内容保存到 2、识别出结果.md 中
with open('Res/2、识别出结果.md', 'w', encoding='utf-8') as f:
    f.write(completion.choices[0].message.content)

print("保存成功！")