【转载】代码：DeepSeek-OCR硅基流动API使用教程：轻松实现高精度OCR识别

import base64 import datetime import re import time import requests from colorama import Fore, Style def deepseek_ocr(image_path): """ 调用硅基流动API进行OCR识别 :param image_path: 图片文件路径 :return: 识别出的文本列表 """ url = "https://api.siliconflow.cn/v1/chat/completions" api_key = "YOUR_API_KEY" # 替换为你的API密钥 model = "deepseek-ai/DeepSeek-OCR" # 构建请求内容 payload = { "model": model, "messages": [ { "role": "user", "content": [ { "type": "image_url", "image_url": { "url": "data:application/pdf;base64," + base64.b64encode( open(image_path, "rb").read()).decode("utf-8") } }, { "type": "text", "text": "<image>\n<|grounding|>OCR this image." } ] } ] } headers = { "Authorization": f"Bearer {api_key}", "Content-Type": "application/json" } # 重试机制，防止网络波动 for i in range(5): try: response = requests.post(url, json=payload, headers=headers) response.raise_for_status() data = response.json() # 检查响应是否包含结果 if "choices" in data and data["choices"]: print(f"{Fore.GREEN}✅ OCR 成功{Style.RESET_ALL}") content = data["choices"][0]["message"]["content"] # 提取所有<|ref|>...<|/ref|>中的文本 texts = re.findall(r'<\|ref\|>(.*?)<\|\/ref\|>', content) return texts else: print(f"{Fore.YELLOW}⚠️ 未获取到识别结果，第 {i+1} 次重试...{Style.RESET_ALL}") time.sleep(1) except Exception as e: print(f"{Fore.RED}❌ 请求异常: {str(e)}，第 {i+1} 次重试...{Style.RESET_ALL}") time.sleep(1) print(f"{Fore.RED}💥 重试 5 次后仍失败，放弃请求{Style.RESET_ALL}") return None if __name__ == "__main__": start_time = datetime.datetime.now() image_path = "your_image.jpg" # 替换为你的图片路径 texts = deepseek_ocr(image_path) end_time = datetime.datetime.now() if texts: print(f"{Fore.GREEN}调用API花费{(end_time - start_time).total_seconds():.2f}秒！\n识别结果：{texts}{Style.RESET_ALL}") else: print(f"{Fore.RED}OCR识别失败{Style.RESET_ALL}")

注：API密钥在硅基流动官网获取：https://cloud.siliconflow.cn/me/account/ak

原文链接：https://blog.csdn.net/qq_58607032/article/details/155609565?fromshare=blogdetail&sharetype=blogdetail&sharerId=155609565&sharerefer=PC&sharesource=Autumn_horse&sharefrom=from_link