\u4f60\u6709\u6ca1\u6709\u9047\u5230\u8fc7\u8fd9\u6837\u7684\u573a\u666f？\u60f3\u5728\u4f1a\u8bae\u5ba4\u3001\u56fe\u4e66\u9986\u6216\u8005\u751f\u4ea7\u7ebf\u4e0a\u90e8\u7f72\u4e00\u4e2a\u624b\u673a\u68c0\u6d4b\u7cfb\u7edf，\u7528\u6765\u7edf\u8ba1\u624b\u673a\u4f7f\u7528\u60c5\u51b5\u6216\u8005\u786e\u4fdd\u7279\u5b9a\u533a\u57df\u7684\u5b89\u5168，\u7ed3\u679c\u53d1\u73b0\u670d\u52a1\u5668\u914d\u7f6e\u592a\u4f4e，\u8dd1\u4e0d\u52a8\u90a3\u4e9b\u52a8\u8f84\u51e0\u4e2aG\u7684\u68c0\u6d4b\u6a21\u578b\u3002<\/p>\n
\u6211\u4e4b\u524d\u5c31\u78b0\u5230\u8fc7\u8fd9\u4e2a\u95ee\u9898\u3002\u5ba2\u6237\u6709\u4e00\u53f0\u8001\u65e7\u7684\u670d\u52a1\u5668，\u5185\u5b58\u53ea\u67094GB，CPU\u4e5f\u662f\u597d\u51e0\u5e74\u524d\u7684\u578b\u53f7，\u4f46\u9700\u8981\u5728\u5165\u53e3\u5904\u5b9e\u65f6\u68c0\u6d4b\u662f\u5426\u6709\u4eba\u643a\u5e26\u624b\u673a\u8fdb\u5165\u3002\u5e02\u9762\u4e0a\u7684\u4e3b\u6d41\u68c0\u6d4b\u6a21\u578b\u8981\u4e48\u592a\u5927\u8dd1\u4e0d\u8d77\u6765，\u8981\u4e48\u901f\u5ea6\u592a\u6162\u8fbe\u4e0d\u5230\u5b9e\u65f6\u8981\u6c42\u3002<\/p>\n
\u76f4\u5230\u6211\u53d1\u73b0\u4e86\u963f\u91cc\u5df4\u5df4\u5f00\u6e90\u7684DAMO-YOLO\u624b\u673a\u68c0\u6d4b\u6a21\u578b\u3002\u8fd9\u4e2a\u6a21\u578b\u53ea\u6709125MB\u5927\u5c0f，\u5728\u6807\u51c6\u6d4b\u8bd5\u96c6\u4e0a\u80fd\u8fbe\u523088.8%\u7684\u51c6\u786e\u7387，\u5355\u5f20\u56fe\u7247\u63a8\u7406\u65f6\u95f4\u53ea\u9700\u89813.83\u6beb\u79d2\u3002\u66f4\u91cd\u8981\u7684\u662f，\u5b83\u4e13\u95e8\u9488\u5bf9\u624b\u673a\u8fd9\u4e00\u5355\u4e00\u7c7b\u522b\u8fdb\u884c\u4e86\u4f18\u5316，\u5728\u4fdd\u6301\u9ad8\u7cbe\u5ea6\u7684\u540c\u65f6\u5927\u5e45\u51cf\u5c0f\u4e86\u6a21\u578b\u4f53\u79ef\u3002<\/p>\n
\u4eca\u5929\u6211\u5c31\u6765\u5206\u4eab\u5982\u4f55\u628a\u8fd9\u4e2a\u8f7b\u91cf\u7ea7\u4f46\u9ad8\u6027\u80fd\u7684\u624b\u673a\u68c0\u6d4b\u6a21\u578b\u90e8\u7f72\u5230\u4f4e\u914d\u7f6e\u670d\u52a1\u5668\u4e0a，\u8ba9\u4f60\u5373\u4f7f\u5728\u6ca1\u6709\u9ad8\u7aefGPU\u7684\u673a\u5668\u4e0a\u4e5f\u80fd\u5b9e\u73b0\u5b9e\u65f6\u624b\u673a\u68c0\u6d4b\u3002<\/p>\n
\u4f60\u53ef\u80fd\u542c\u8bf4\u8fc7YOLO\u7cfb\u5217\u7684\u76ee\u6807\u68c0\u6d4b\u6a21\u578b，\u4eceYOLOv1\u5230\u6700\u65b0\u7684YOLOv11，\u6a21\u578b\u6027\u80fd\u8d8a\u6765\u8d8a\u5f3a，\u4f46\u4f53\u79ef\u4e5f\u8d8a\u6765\u8d8a\u5927\u3002\u5bf9\u4e8e\u5f88\u591a\u5b9e\u9645\u5e94\u7528\u573a\u666f\u6765\u8bf4，\u6211\u4eec\u5e76\u4e0d\u9700\u8981\u68c0\u6d4b\u6210\u767e\u4e0a\u5343\u4e2a\u7c7b\u522b，\u53ef\u80fd\u53ea\u9700\u8981\u68c0\u6d4b\u4e00\u4e24\u4e2a\u7279\u5b9a\u7684\u7269\u4f53\u3002<\/p>\n
DAMO-YOLO\u624b\u673a\u68c0\u6d4b\u6a21\u578b\u5c31\u662f\u9488\u5bf9\u8fd9\u79cd\u9700\u6c42\u8bbe\u8ba1\u7684\u3002\u5b83\u53ea\u4e13\u6ce8\u4e8e\u68c0\u6d4b\u624b\u673a\u8fd9\u4e00\u4e2a\u7c7b\u522b，\u6240\u4ee5\u6a21\u578b\u7ed3\u6784\u53ef\u4ee5\u505a\u5f97\u66f4\u52a0\u7cbe\u7b80\u3002125MB\u7684\u4f53\u79ef\u610f\u5473\u7740：<\/p>\n
\u8fd9\u4e2a\u6a21\u578b\u5728COCO\u6570\u636e\u96c6\u4e0a\u7684AP@0.5\u8fbe\u5230\u4e8688.8%\u3002AP@0.5\u662f\u4ec0\u4e48\u610f\u601d\u5462？\u7b80\u5355\u6765\u8bf4，\u5c31\u662f\u5f53\u68c0\u6d4b\u6846\u4e0e\u771f\u5b9e\u6846\u7684\u91cd\u53e0\u9762\u79ef\u8d85\u8fc750%\u65f6，\u6a21\u578b\u670988.8%\u7684\u6982\u7387\u80fd\u6b63\u786e\u8bc6\u522b\u51fa\u624b\u673a\u3002<\/p>\n
3.83\u6beb\u79d2\u7684\u63a8\u7406\u901f\u5ea6\u53c8\u662f\u4ec0\u4e48\u6982\u5ff5？\u8fd9\u610f\u5473\u7740\u5728\u4e00\u79d2\u949f\u5185，\u8fd9\u4e2a\u6a21\u578b\u53ef\u4ee5\u5904\u7406\u5927\u7ea6260\u5f20\u56fe\u7247\u3002\u5bf9\u4e8e\u5927\u591a\u6570\u5b9e\u65f6\u76d1\u63a7\u573a\u666f（\u901a\u5e38\u6bcf\u79d225-30\u5e27），\u8fd9\u4e2a\u901f\u5ea6\u7ef0\u7ef0\u6709\u4f59\u3002<\/p>\n
\u66f4\u91cd\u8981\u7684\u662f，\u8fd9\u4e2a\u901f\u5ea6\u662f\u5728T4\u663e\u5361\u4e0a\u6d4b\u5f97\u7684\u3002\u5728\u5b9e\u9645\u7684\u4f4e\u914d\u670d\u52a1\u5668\u4e0a，\u5373\u4f7f\u53ea\u7528CPU，\u4e5f\u80fd\u8fbe\u5230\u6bcf\u79d2\u51e0\u5341\u5e27\u7684\u5904\u7406\u901f\u5ea6，\u5b8c\u5168\u6ee1\u8db3\u5b9e\u65f6\u68c0\u6d4b\u7684\u9700\u6c42\u3002<\/p>\n
\u5728\u5f00\u59cb\u90e8\u7f72\u4e4b\u524d，\u6211\u4eec\u5148\u786e\u8ba4\u4e00\u4e0b\u4f60\u7684\u670d\u52a1\u5668\u662f\u5426\u6ee1\u8db3\u57fa\u672c\u8981\u6c42：<\/p>\n
\u5982\u679c\u4f60\u7528\u7684\u662fWindows\u670d\u52a1\u5668，\u5efa\u8bae\u5148\u5b89\u88c5WSL2（Windows Subsystem for Linux），\u7136\u540e\u5728WSL2\u4e2d\u8fd0\u884c\u3002\u4e0d\u8fc7\u4ece\u6211\u7684\u7ecf\u9a8c\u6765\u770b，Linux\u73af\u5883\u4e0b\u7684\u90e8\u7f72\u4f1a\u66f4\u7b80\u5355\u7a33\u5b9a\u3002<\/p>\n
\u90e8\u7f72\u8fc7\u7a0b\u6bd4\u4f60\u60f3\u7684\u8981\u7b80\u5355\u5f97\u591a\u3002\u6574\u4e2a\u6d41\u7a0b\u53ef\u4ee5\u5206\u4e3a\u4e09\u4e2a\u4e3b\u8981\u6b65\u9aa4：<\/p>\n
\u6211\u4eec\u5148\u4ece\u6700\u57fa\u7840\u7684\u5f00\u59cb\u3002\u6253\u5f00\u4f60\u7684\u670d\u52a1\u5668\u7ec8\u7aef，\u4f9d\u6b21\u6267\u884c\u4ee5\u4e0b\u547d\u4ee4：<\/p>\n
# \u7b2c\u4e00\u6b65：\u8fdb\u5165\u9879\u76ee\u76ee\u5f55（\u5982\u679c\u76ee\u5f55\u4e0d\u5b58\u5728\u4f1a\u81ea\u52a8\u521b\u5efa）
\ncd \/root<\/p>\n
# \u7b2c\u4e8c\u6b65：\u514b\u9686\u9879\u76ee\u4ee3\u7801
\ngit clone https:\/\/github.com\/modelscope\/modelscope.git
\ncd modelscope<\/p>\n
# \u7b2c\u4e09\u6b65：\u5b89\u88c5ModelScope\u6846\u67b6
\npip install modelscope -i https:\/\/mirrors.aliyun.com\/pypi\/simple\/<\/p>\n
# \u7b2c\u56db\u6b65：\u8fdb\u5165\u624b\u673a\u68c0\u6d4b\u9879\u76ee\u76ee\u5f55
\ncd \/root\/cv_tinynas_object-detection_damoyolo_phone<\/p>\n
\u5982\u679c\u4e00\u5207\u987a\u5229，\u4f60\u5e94\u8be5\u80fd\u770b\u5230\u9879\u76ee\u76ee\u5f55\u4e0b\u6709\u51e0\u4e2a\u5173\u952e\u6587\u4ef6：<\/p>\n
\u63a5\u4e0b\u6765\u5b89\u88c5\u8fd0\u884c\u6240\u9700\u7684\u6240\u6709\u8f6f\u4ef6\u5305：<\/p>\n
# \u5b89\u88c5\u9879\u76ee\u4f9d\u8d56
\npip install -r requirements.txt -i https:\/\/mirrors.aliyun.com\/pypi\/simple\/<\/p>\n
\u8fd9\u91cc\u6709\u51e0\u4e2a\u5173\u952e\u4f9d\u8d56\u9700\u8981\u7279\u522b\u8bf4\u660e：<\/p>\n
\u5b89\u88c5\u8fc7\u7a0b\u4e2d\u5982\u679c\u9047\u5230\u7f51\u7edc\u95ee\u9898，\u53ef\u4ee5\u5c1d\u8bd5\u66f4\u6362pip\u6e90\u3002\u4e0a\u9762\u547d\u4ee4\u4e2d\u4f7f\u7528\u7684-i https:\/\/mirrors.aliyun.com\/pypi\/simple\/\u5c31\u662f\u963f\u91cc\u4e91\u7684\u955c\u50cf\u6e90，\u4e0b\u8f7d\u901f\u5ea6\u4f1a\u5feb\u5f88\u591a\u3002<\/p>\n
\u8fd9\u662f\u6700\u7b80\u5355\u76f4\u89c2\u7684\u4f7f\u7528\u65b9\u5f0f\u3002\u542f\u52a8\u670d\u52a1\u540e，\u4f60\u4f1a\u5f97\u5230\u4e00\u4e2a\u7f51\u9875\u5730\u5740，\u5728\u6d4f\u89c8\u5668\u4e2d\u6253\u5f00\u5c31\u80fd\u76f4\u63a5\u4f7f\u7528\u3002<\/p>\n
\u542f\u52a8\u670d\u52a1\u7684\u547d\u4ee4\u5f88\u7b80\u5355：<\/p>\n
# \u65b9\u6cd5\u4e00：\u4f7f\u7528\u542f\u52a8\u811a\u672c（\u63a8\u8350）
\n.\/start.sh<\/p>\n
# \u65b9\u6cd5\u4e8c：\u76f4\u63a5\u8fd0\u884cPython\u7a0b\u5e8f
\npython3 app.py<\/p>\n
\u6267\u884c\u540e，\u4f60\u4f1a\u770b\u5230\u7c7b\u4f3c\u8fd9\u6837\u7684\u8f93\u51fa：<\/p>\n
Running on local URL: http:\/\/0.0.0.0:7860
\nRunning on public URL: https:\/\/xxxx.gradio.live<\/p>\n
\u73b0\u5728\u6253\u5f00\u6d4f\u89c8\u5668，\u8bbf\u95eehttp:\/\/\u4f60\u7684\u670d\u52a1\u5668IP:7860（\u5982\u679c\u662f\u672c\u5730\u670d\u52a1\u5668，\u5c31\u662fhttp:\/\/localhost:7860）\u3002<\/p>\n
\u4f60\u4f1a\u770b\u5230\u4e00\u4e2a\u7b80\u6d01\u7684Web\u754c\u9762，\u4e3b\u8981\u5305\u542b\u4ee5\u4e0b\u51e0\u4e2a\u90e8\u5206：<\/p>\n
\u4f7f\u7528\u6d41\u7a0b\u4e5f\u5f88\u76f4\u89c2：<\/p>\n
\u68c0\u6d4b\u7ed3\u679c\u4f1a\u7528\u7ea2\u8272\u7684\u65b9\u6846\u6807\u51fa\u56fe\u7247\u4e2d\u6240\u6709\u7684\u624b\u673a，\u6bcf\u4e2a\u65b9\u6846\u65c1\u8fb9\u8fd8\u4f1a\u663e\u793a\u4e00\u4e2a\u7f6e\u4fe1\u5ea6\u5206\u6570，\u8868\u793a\u6a21\u578b\u5bf9\u8fd9\u4e2a\u68c0\u6d4b\u7ed3\u679c\u7684\u628a\u63e1\u7a0b\u5ea6\u3002\u5206\u6570\u8d8a\u9ad8，\u8bf4\u660e\u6a21\u578b\u8d8a\u786e\u5b9a\u8fd9\u91cc\u6709\u4e2a\u624b\u673a\u3002<\/p>\n
\u5982\u679c\u4f60\u9700\u8981\u5728\u5176\u4ed6\u7a0b\u5e8f\u4e2d\u96c6\u6210\u624b\u673a\u68c0\u6d4b\u529f\u80fd，\u6216\u8005\u60f3\u8981\u6279\u91cf\u5904\u7406\u56fe\u7247，\u4f7f\u7528Python API\u4f1a\u66f4\u52a0\u65b9\u4fbf\u3002<\/p>\n
\u4e0b\u9762\u662f\u4e00\u4e2a\u5b8c\u6574\u7684\u793a\u4f8b\u4ee3\u7801：<\/p>\n
# \u5bfc\u5165\u5fc5\u8981\u7684\u5e93
\nfrom modelscope.pipelines import pipeline
\nfrom modelscope.utils.constant import Tasks
\nimport cv2
\nimport matplotlib.pyplot as plt<\/p>\n
def detect_phones_in_image(image_path):
\n """
\n \u68c0\u6d4b\u56fe\u7247\u4e2d\u7684\u624b\u673a<\/p>\n
\u53c2\u6570:
\n image_path: \u56fe\u7247\u6587\u4ef6\u8def\u5f84<\/p>\n
\u8fd4\u56de:
\n \u68c0\u6d4b\u7ed3\u679c，\u5305\u542b\u8fb9\u754c\u6846\u548c\u7f6e\u4fe1\u5ea6
\n """
\n # \u7b2c\u4e00\u6b65：\u521b\u5efa\u68c0\u6d4b\u5668
\n # \u8fd9\u91cc\u6307\u5b9a\u4efb\u52a1\u7c7b\u578b\u4e3a\u7279\u5b9a\u9886\u57df\u76ee\u6807\u68c0\u6d4b
\n detector = pipeline(
\n Tasks.domain_specific_object_detection,
\n model='damo\/cv_tinynas_object-detection_damoyolo_phone',
\n cache_dir='\/root\/ai-models', # \u6a21\u578b\u7f13\u5b58\u76ee\u5f55
\n trust_remote_code=True # \u4fe1\u4efb\u8fdc\u7a0b\u4ee3\u7801（\u5fc5\u987b\u8bbe\u7f6e）
\n )<\/p>\n
# \u7b2c\u4e8c\u6b65：\u6267\u884c\u68c0\u6d4b
\n result = detector(image_path)<\/p>\n
# \u7b2c\u4e09\u6b65：\u5904\u7406\u7ed3\u679c
\n print("\u68c0\u6d4b\u5b8c\u6210！")
\n print(f"\u5171\u68c0\u6d4b\u5230 {len(result['boxes'])} \u4e2a\u624b\u673a")<\/p>\n
# \u663e\u793a\u6bcf\u4e2a\u68c0\u6d4b\u7ed3\u679c\u7684\u8be6\u7ec6\u4fe1\u606f
\n for i, (box, score) in enumerate(zip(result['boxes'], result['scores'])):
\n print(f"\u624b\u673a {i+1}:")
\n print(f" \u4f4d\u7f6e: \u5de6\u4e0a({box[0]:.1f}, {box[1]:.1f}), "
\n f"\u53f3\u4e0b({box[2]:.1f}, {box[3]:.1f})")
\n print(f" \u7f6e\u4fe1\u5ea6: {score:.3f}")<\/p>\n
return result<\/p>\n
# \u4f7f\u7528\u793a\u4f8b
\nif __name__ == "__main__":
\n # \u68c0\u6d4b\u5355\u5f20\u56fe\u7247
\n result = detect_phones_in_image("test_image.jpg")<\/p>\n
# \u5982\u679c\u4f60\u60f3\u8981\u53ef\u89c6\u5316\u7ed3\u679c
\n image = cv2.imread("test_image.jpg")
\n image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)<\/p>\n
# \u5728\u56fe\u7247\u4e0a\u7ed8\u5236\u68c0\u6d4b\u6846
\n for box, score in zip(result['boxes'], result['scores']):
\n x1, y1, x2, y2 = map(int, box)
\n # \u7ed8\u5236\u77e9\u5f62\u6846
\n cv2.rectangle(image, (x1, y1), (x2, y2), (255, 0, 0), 2)
\n # \u6dfb\u52a0\u7f6e\u4fe1\u5ea6\u6587\u672c
\n cv2.putText(image, f"phone: {score:.2f}",
\n (x1, y1-10), cv2.FONT_HERSHEY_SIMPLEX,
\n 0.5, (255, 0, 0), 2)<\/p>\n
# \u4fdd\u5b58\u7ed3\u679c\u56fe\u7247
\n cv2.imwrite("result.jpg", cv2.cvtColor(image, cv2.COLOR_RGB2BGR))
\n print("\u7ed3\u679c\u5df2\u4fdd\u5b58\u5230 result.jpg")<\/p>\n
\u8fd9\u6bb5\u4ee3\u7801\u505a\u4e86\u51e0\u4ef6\u4e8b\u60c5：<\/p>\n
\u5728\u5b9e\u9645\u5e94\u7528\u4e2d，\u6211\u4eec\u7ecf\u5e38\u9700\u8981\u5904\u7406\u5927\u91cf\u56fe\u7247\u3002\u4e0b\u9762\u662f\u4e00\u4e2a\u6279\u91cf\u5904\u7406\u7684\u793a\u4f8b：<\/p>\n
import os
\nfrom modelscope.pipelines import pipeline
\nfrom modelscope.utils.constant import Tasks<\/p>\n
def batch_detect_phones(image_folder, output_folder):
\n """
\n \u6279\u91cf\u68c0\u6d4b\u6587\u4ef6\u5939\u4e2d\u7684\u6240\u6709\u56fe\u7247<\/p>\n
\u53c2\u6570:
\n image_folder: \u8f93\u5165\u56fe\u7247\u6587\u4ef6\u5939\u8def\u5f84
\n output_folder: \u8f93\u51fa\u7ed3\u679c\u6587\u4ef6\u5939\u8def\u5f84
\n """
\n # \u521b\u5efa\u8f93\u51fa\u6587\u4ef6\u5939
\n os.makedirs(output_folder, exist_ok=True)<\/p>\n
# \u521d\u59cb\u5316\u68c0\u6d4b\u5668（\u53ea\u521d\u59cb\u5316\u4e00\u6b21，\u63d0\u9ad8\u6548\u7387）
\n detector = pipeline(
\n Tasks.domain_specific_object_detection,
\n model='damo\/cv_tinynas_object-detection_damoyolo_phone',
\n cache_dir='\/root\/ai-models',
\n trust_remote_code=True
\n )<\/p>\n
# \u83b7\u53d6\u6240\u6709\u56fe\u7247\u6587\u4ef6
\n image_extensions = ['.jpg', '.jpeg', '.png', '.bmp']
\n image_files = []<\/p>\n
for file in os.listdir(image_folder):
\n if any(file.lower().endswith(ext) for ext in image_extensions):
\n image_files.append(os.path.join(image_folder, file))<\/p>\n
print(f"\u627e\u5230 {len(image_files)} \u5f20\u56fe\u7247\u9700\u8981\u5904\u7406")<\/p>\n
# \u6279\u91cf\u5904\u7406
\n for i, image_path in enumerate(image_files):
\n print(f"\u5904\u7406\u7b2c {i+1}\/{len(image_files)} \u5f20: {os.path.basename(image_path)}")<\/p>\n
try:
\n # \u6267\u884c\u68c0\u6d4b
\n result = detector(image_path)<\/p>\n
# \u4fdd\u5b58\u68c0\u6d4b\u7ed3\u679c\u5230\u6587\u672c\u6587\u4ef6
\n output_file = os.path.join(
\n output_folder,
\n os.path.splitext(os.path.basename(image_path))[0] + ".txt"
\n )<\/p>\n
with open(output_file, 'w') as f:
\n f.write(f"\u56fe\u7247: {os.path.basename(image_path)}\\\\n")
\n f.write(f"\u68c0\u6d4b\u5230\u624b\u673a\u6570\u91cf: {len(result['boxes'])}\\\\n\\\\n")<\/p>\n
for j, (box, score) in enumerate(zip(result['boxes'], result['scores'])):
\n f.write(f"\u624b\u673a {j+1}:\\\\n")
\n f.write(f" \u8fb9\u754c\u6846: [{box[0]:.1f}, {box[1]:.1f}, {box[2]:.1f}, {box[3]:.1f}]\\\\n")
\n f.write(f" \u7f6e\u4fe1\u5ea6: {score:.4f}\\\\n")
\n f.write("-" * 40 + "\\\\n")<\/p>\n
print(f" \u7ed3\u679c\u5df2\u4fdd\u5b58\u5230: {output_file}")<\/p>\n
except Exception as e:
\n print(f" \u5904\u7406\u5931\u8d25: {str(e)}")<\/p>\n
print("\u6279\u91cf\u5904\u7406\u5b8c\u6210！")<\/p>\n
# \u4f7f\u7528\u793a\u4f8b
\nif __name__ == "__main__":
\n # \u5904\u7406images\u6587\u4ef6\u5939\u4e2d\u7684\u6240\u6709\u56fe\u7247，\u7ed3\u679c\u4fdd\u5b58\u5230results\u6587\u4ef6\u5939
\n batch_detect_phones("images", "results")<\/p>\n
\u5728\u5185\u5b58\u6709\u9650\u7684\u670d\u52a1\u5668\u4e0a\u8fd0\u884c\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b，\u5185\u5b58\u7ba1\u7406\u7279\u522b\u91cd\u8981\u3002\u4e0b\u9762\u662f\u4e00\u4e9b\u5b9e\u7528\u7684\u4f18\u5316\u6280\u5de7：<\/p>\n
import gc
\nimport torch<\/p>\n
def memory_efficient_detection(image_path):
\n """
\n \u5185\u5b58\u53cb\u597d\u7684\u68c0\u6d4b\u51fd\u6570<\/p>\n
\u8fd9\u4e2a\u51fd\u6570\u4f1a\u5728\u68c0\u6d4b\u5b8c\u6210\u540e\u7acb\u5373\u91ca\u653e\u5185\u5b58，
\n \u9002\u5408\u5728\u5185\u5b58\u6709\u9650\u7684\u8bbe\u5907\u4e0a\u957f\u65f6\u95f4\u8fd0\u884c\u3002
\n """
\n # \u53ea\u5728\u9700\u8981\u65f6\u52a0\u8f7d\u6a21\u578b
\n from modelscope.pipelines import pipeline
\n from modelscope.utils.constant import Tasks<\/p>\n
try:
\n # \u521b\u5efa\u68c0\u6d4b\u5668
\n detector = pipeline(
\n Tasks.domain_specific_object_detection,
\n model='damo\/cv_tinynas_object-detection_damoyolo_phone',
\n cache_dir='\/root\/ai-models',
\n trust_remote_code=True
\n )<\/p>\n
# \u6267\u884c\u68c0\u6d4b
\n result = detector(image_path)<\/p>\n
# \u7acb\u5373\u5220\u9664\u68c0\u6d4b\u5668\u91ca\u653e\u5185\u5b58
\n del detector<\/p>\n
# \u5f3a\u5236\u5783\u573e\u56de\u6536
\n gc.collect()<\/p>\n
# \u5982\u679c\u4f7f\u7528GPU，\u6e05\u7a7a\u7f13\u5b58
\n if torch.cuda.is_available():
\n torch.cuda.empty_cache()<\/p>\n
return result<\/p>\n
except Exception as e:
\n print(f"\u68c0\u6d4b\u5931\u8d25: {str(e)}")
\n return None<\/p>\n
\u5904\u7406\u5927\u5c3a\u5bf8\u56fe\u7247\u4f1a\u6d88\u8017\u66f4\u591a\u5185\u5b58\u3002\u6211\u4eec\u53ef\u4ee5\u5148\u8c03\u6574\u56fe\u7247\u5927\u5c0f，\u518d\u8fdb\u884c\u68c0\u6d4b：<\/p>\n
from PIL import Image
\nimport numpy as np<\/p>\n
def resize_and_detect(image_path, max_size=1024):
\n """
\n \u5148\u8c03\u6574\u56fe\u7247\u5927\u5c0f\u518d\u68c0\u6d4b，\u51cf\u5c11\u5185\u5b58\u4f7f\u7528<\/p>\n
\u53c2\u6570:
\n image_path: \u56fe\u7247\u8def\u5f84
\n max_size: \u56fe\u7247\u6700\u5927\u8fb9\u957f，\u9ed8\u8ba41024\u50cf\u7d20
\n """
\n # \u4f7f\u7528PIL\u6253\u5f00\u56fe\u7247\u5e76\u8c03\u6574\u5927\u5c0f
\n img = Image.open(image_path)<\/p>\n
# \u83b7\u53d6\u539f\u59cb\u5c3a\u5bf8
\n width, height = img.size<\/p>\n
# \u8ba1\u7b97\u8c03\u6574\u540e\u7684\u5c3a\u5bf8（\u4fdd\u6301\u5bbd\u9ad8\u6bd4）
\n if max(width, height) > max_size:
\n if width > height:
\n new_width = max_size
\n new_height = int(height * (max_size \/ width))
\n else:
\n new_height = max_size
\n new_width = int(width * (max_size \/ height))<\/p>\n
img = img.resize((new_width, new_height), Image.Resampling.LANCZOS)
\n print(f"\u56fe\u7247\u4ece {width}x{height} \u8c03\u6574\u5230 {new_width}x{new_height}")<\/p>\n
# \u4fdd\u5b58\u8c03\u6574\u540e\u7684\u4e34\u65f6\u6587\u4ef6
\n temp_path = "temp_resized.jpg"
\n img.save(temp_path)<\/p>\n
# \u4f7f\u7528\u8c03\u6574\u540e\u7684\u56fe\u7247\u8fdb\u884c\u68c0\u6d4b
\n from modelscope.pipelines import pipeline
\n from modelscope.utils.constant import Tasks<\/p>\n
detector = pipeline(
\n Tasks.domain_specific_object_detection,
\n model='damo\/cv_tinynas_object-detection_damoyolo_phone',
\n cache_dir='\/root\/ai-models',
\n trust_remote_code=True
\n )<\/p>\n
result = detector(temp_path)<\/p>\n
# \u6e05\u7406\u4e34\u65f6\u6587\u4ef6
\n import os
\n os.remove(temp_path)<\/p>\n
return result<\/p>\n
\u5bf9\u4e8e\u957f\u671f\u8fd0\u884c\u7684\u670d\u52a1，\u6211\u4eec\u9700\u8981\u4e00\u4e2a\u53ef\u9760\u7684\u7ba1\u7406\u65b9\u5f0f\u3002\u4e0b\u9762\u662f\u4e00\u4e2a\u7b80\u5355\u7684\u670d\u52a1\u7ba1\u7406\u811a\u672c：<\/p>\n
#!\/bin\/bash
\n# service_manager.sh – \u624b\u673a\u68c0\u6d4b\u670d\u52a1\u7ba1\u7406\u811a\u672c<\/p>\n
SERVICE_DIR="\/root\/cv_tinynas_object-detection_damoyolo_phone"
\nLOG_FILE="$SERVICE_DIR\/service.log"
\nPID_FILE="$SERVICE_DIR\/service.pid"<\/p>\n
start_service() {
\n echo "\u6b63\u5728\u542f\u52a8\u624b\u673a\u68c0\u6d4b\u670d\u52a1…"
\n cd $SERVICE_DIR
\n nohup python3 app.py > $LOG_FILE 2>&1 &
\n echo $! > $PID_FILE
\n echo "\u670d\u52a1\u5df2\u542f\u52a8，PID: $(cat $PID_FILE)"
\n echo "\u65e5\u5fd7\u6587\u4ef6: $LOG_FILE"
\n echo "Web\u754c\u9762: http:\/\/localhost:7860"
\n}<\/p>\n
stop_service() {
\n if [ -f $PID_FILE ]; then
\n PID=$(cat $PID_FILE)
\n echo "\u6b63\u5728\u505c\u6b62\u670d\u52a1 (PID: $PID)…"
\n kill $PID
\n rm $PID_FILE
\n echo "\u670d\u52a1\u5df2\u505c\u6b62"
\n else
\n echo "\u670d\u52a1\u672a\u8fd0\u884c"
\n fi
\n}<\/p>\n
restart_service() {
\n stop_service
\n sleep 2
\n start_service
\n}<\/p>\n
check_status() {
\n if [ -f $PID_FILE ]; then
\n PID=$(cat $PID_FILE)
\n if ps -p $PID > \/dev\/null; then
\n echo "\u670d\u52a1\u6b63\u5728\u8fd0\u884c (PID: $PID)"
\n echo "\u67e5\u770b\u65e5\u5fd7: tail -f $LOG_FILE"
\n else
\n echo "\u670d\u52a1\u8fdb\u7a0b\u4e0d\u5b58\u5728，\u4f46PID\u6587\u4ef6\u5b58\u5728"
\n rm $PID_FILE
\n fi
\n else
\n echo "\u670d\u52a1\u672a\u8fd0\u884c"
\n fi
\n}<\/p>\n
case "$1" in
\n start)
\n start_service
\n ;;
\n stop)
\n stop_service
\n ;;
\n restart)
\n restart_service
\n ;;
\n status)
\n check_status
\n ;;
\n *)
\n echo "\u4f7f\u7528\u65b9\u6cd5: $0 {start|stop|restart|status}"
\n exit 1
\n ;;
\nesac<\/p>\n
\u4f7f\u7528\u65b9\u6cd5：<\/p>\n
# \u7ed9\u811a\u672c\u6dfb\u52a0\u6267\u884c\u6743\u9650
\nchmod +x service_manager.sh<\/p>\n
# \u542f\u52a8\u670d\u52a1
\n.\/service_manager.sh start<\/p>\n
# \u67e5\u770b\u72b6\u6001
\n.\/service_manager.sh status<\/p>\n
# \u505c\u6b62\u670d\u52a1
\n.\/service_manager.sh stop<\/p>\n
# \u91cd\u542f\u670d\u52a1
\n.\/service_manager.sh restart<\/p>\n
\u6211\u6700\u8fd1\u5728\u4e00\u4e2a\u5ba2\u6237\u90a3\u91cc\u90e8\u7f72\u4e86\u8fd9\u4e2a\u7cfb\u7edf，\u7528\u6765\u7edf\u8ba1\u4f1a\u8bae\u5ba4\u4e2d\u624b\u673a\u7684\u4f7f\u7528\u60c5\u51b5\u3002\u4ed6\u4eec\u5728\u4f1a\u8bae\u5ba4\u5165\u53e3\u5b89\u88c5\u4e86\u6444\u50cf\u5934，\u901a\u8fc7\u6211\u4eec\u7684\u68c0\u6d4b\u7cfb\u7edf\u5b9e\u65f6\u5206\u6790：<\/p>\n
import time
\nfrom datetime import datetime<\/p>\n
class MeetingRoomMonitor:
\n """\u4f1a\u8bae\u5ba4\u624b\u673a\u4f7f\u7528\u76d1\u63a7\u5668"""<\/p>\n
def __init__(self):
\n self.detector = None
\n self.phone_count_history = []<\/p>\n
def initialize_detector(self):
\n """\u521d\u59cb\u5316\u68c0\u6d4b\u5668"""
\n from modelscope.pipelines import pipeline
\n from modelscope.utils.constant import Tasks<\/p>\n
self.detector = pipeline(
\n Tasks.domain_specific_object_detection,
\n model='damo\/cv_tinynas_object-detection_damoyolo_phone',
\n cache_dir='\/root\/ai-models',
\n trust_remote_code=True
\n )<\/p>\n
def monitor_meeting_room(self, duration_minutes=60, interval_seconds=10):
\n """
\n \u76d1\u63a7\u4f1a\u8bae\u5ba4\u4e00\u6bb5\u65f6\u95f4<\/p>\n
\u53c2\u6570:
\n duration_minutes: \u76d1\u63a7\u65f6\u957f（\u5206\u949f）
\n interval_seconds: \u68c0\u6d4b\u95f4\u9694（\u79d2）
\n """
\n if not self.detector:
\n self.initialize_detector()<\/p>\n
total_checks = (duration_minutes * 60) \/\/ interval_seconds
\n print(f"\u5f00\u59cb\u76d1\u63a7，\u5c06\u6301\u7eed {duration_minutes} \u5206\u949f，\u6bcf {interval_seconds} \u79d2\u68c0\u6d4b\u4e00\u6b21")<\/p>\n
for i in range(total_checks):
\n # \u8fd9\u91cc\u5e94\u8be5\u662f\u4ece\u6444\u50cf\u5934\u83b7\u53d6\u56fe\u7247，\u793a\u4f8b\u4e2d\u4f7f\u7528\u9759\u6001\u56fe\u7247
\n image_path = "meeting_room_snapshot.jpg"<\/p>\n
try:
\n result = self.detector(image_path)
\n phone_count = len(result['boxes'])<\/p>\n
# \u8bb0\u5f55\u7ed3\u679c
\n timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
\n self.phone_count_history.append({
\n 'time': timestamp,
\n 'count': phone_count
\n })<\/p>\n
print(f"[{timestamp}] \u68c0\u6d4b\u5230 {phone_count} \u90e8\u624b\u673a")<\/p>\n
# \u5982\u679c\u624b\u673a\u6570\u91cf\u8d85\u8fc7\u9608\u503c，\u53d1\u51fa\u63d0\u9192
\n if phone_count > 5: # \u5047\u8bbe\u4f1a\u8bae\u5ba4\u6700\u591a\u5141\u8bb85\u90e8\u624b\u673a
\n self.send_alert(phone_count)<\/p>\n
except Exception as e:
\n print(f"\u68c0\u6d4b\u5931\u8d25: {str(e)}")<\/p>\n
# \u7b49\u5f85\u4e0b\u4e00\u6b21\u68c0\u6d4b
\n time.sleep(interval_seconds)<\/p>\n
# \u751f\u6210\u62a5\u544a
\n self.generate_report()<\/p>\n
def send_alert(self, phone_count):
\n """\u53d1\u9001\u63d0\u9192"""
\n print(f"\u26a0\ufe0f \u8b66\u544a: \u68c0\u6d4b\u5230 {phone_count} \u90e8\u624b\u673a，\u8d85\u8fc7\u9650\u5236！")
\n # \u8fd9\u91cc\u53ef\u4ee5\u6dfb\u52a0\u90ae\u4ef6\u3001\u77ed\u4fe1\u7b49\u901a\u77e5\u903b\u8f91<\/p>\n
def generate_report(self):
\n """\u751f\u6210\u76d1\u63a7\u62a5\u544a"""
\n if not self.phone_count_history:
\n print("\u6ca1\u6709\u68c0\u6d4b\u6570\u636e")
\n return<\/p>\n
total_detections = len(self.phone_count_history)
\n max_phones = max(item['count'] for item in self.phone_count_history)
\n avg_phones = sum(item['count'] for item in self.phone_count_history) \/ total_detections<\/p>\n
print("\\\\n" + "="*50)
\n print("\u4f1a\u8bae\u5ba4\u624b\u673a\u4f7f\u7528\u7edf\u8ba1\u62a5\u544a")
\n print("="*50)
\n print(f"\u76d1\u63a7\u65f6\u6bb5: {self.phone_count_history[0]['time']} \u5230 {self.phone_count_history[-1]['time']}")
\n print(f"\u68c0\u6d4b\u6b21\u6570: {total_detections}")
\n print(f"\u6700\u5927\u540c\u65f6\u4f7f\u7528\u624b\u673a\u6570: {max_phones}")
\n print(f"\u5e73\u5747\u624b\u673a\u4f7f\u7528\u6570: {avg_phones:.1f}")
\n print("\\\\n\u65f6\u95f4\u7ebf:")<\/p>\n
for item in self.phone_count_history[-10:]: # \u663e\u793a\u6700\u540e10\u6b21\u68c0\u6d4b
\n print(f" {item['time']}: {item['count']} \u90e8\u624b\u673a")<\/p>\n
\u5728\u6709\u4e9b\u751f\u4ea7\u73af\u5883\u4e2d，\u4e0d\u5141\u8bb8\u643a\u5e26\u624b\u673a\u8fdb\u5165\u3002\u8fd9\u4e2a\u7cfb\u7edf\u53ef\u4ee5\u5b9e\u65f6\u68c0\u6d4b\u662f\u5426\u6709\u4eba\u8fdd\u89c4\u643a\u5e26\u624b\u673a：<\/p>\n
class ProductionLineMonitor:
\n """\u751f\u4ea7\u7ebf\u624b\u673a\u68c0\u6d4b\u76d1\u63a7"""<\/p>\n
def __init__(self, camera_urls):
\n """
\n \u521d\u59cb\u5316\u76d1\u63a7\u5668<\/p>\n
\u53c2\u6570:
\n camera_urls: \u6444\u50cf\u5934URL\u5217\u8868
\n """
\n self.camera_urls = camera_urls
\n self.detector = None
\n self.violation_records = []<\/p>\n
def check_single_frame(self, frame):
\n """\u68c0\u67e5\u5355\u5e27\u56fe\u7247"""
\n if not self.detector:
\n self.initialize_detector()<\/p>\n
result = self.detector(frame)
\n phones_detected = len(result['boxes']) > 0<\/p>\n
if phones_detected:
\n # \u8bb0\u5f55\u8fdd\u89c4\u4fe1\u606f
\n violation_info = {
\n 'time': datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
\n 'phone_count': len(result['boxes']),
\n 'confidence_scores': [float(score) for score in result['scores']],
\n 'frame': frame # \u5b9e\u9645\u5e94\u7528\u4e2d\u53ef\u80fd\u53ea\u4fdd\u5b58\u8def\u5f84\u6216\u7f29\u7565\u56fe
\n }
\n self.violation_records.append(violation_info)<\/p>\n
# \u4fdd\u5b58\u8bc1\u636e\u56fe\u7247
\n self.save_evidence(frame, result)<\/p>\n
return True, result
\n else:
\n return False, None<\/p>\n
def save_evidence(self, frame, result):
\n """\u4fdd\u5b58\u8fdd\u89c4\u8bc1\u636e"""
\n import cv2<\/p>\n
# \u5728\u56fe\u7247\u4e0a\u7ed8\u5236\u68c0\u6d4b\u6846
\n for box, score in zip(result['boxes'], result['scores']):
\n x1, y1, x2, y2 = map(int, box)
\n cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 0, 255), 3)
\n cv2.putText(frame, f"Phone: {score:.2f}",
\n (x1, y1-10), cv2.FONT_HERSHEY_SIMPLEX,
\n 1, (0, 0, 255), 2)<\/p>\n
# \u4fdd\u5b58\u56fe\u7247
\n timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
\n filename = f"violation_{timestamp}.jpg"
\n cv2.imwrite(filename, frame)
\n print(f"\u8fdd\u89c4\u8bc1\u636e\u5df2\u4fdd\u5b58: {filename}")<\/p>\n
\u6211\u5728\u51e0\u53f0\u4e0d\u540c\u914d\u7f6e\u7684\u670d\u52a1\u5668\u4e0a\u6d4b\u8bd5\u4e86\u8fd9\u4e2a\u6a21\u578b：<\/p>\n
\u6d4b\u8bd5\u73af\u58831：\u4f4e\u914d\u4e91\u670d\u52a1\u5668<\/p>\n
\u6d4b\u8bd5\u73af\u58832：\u8001\u65e7\u529e\u516c\u7535\u8111<\/p>\n
\u6d4b\u8bd5\u73af\u58833：\u6811\u8393\u6d3e4B<\/p>\n
\u4ece\u6d4b\u8bd5\u7ed3\u679c\u53ef\u4ee5\u770b\u51fa，\u5373\u4f7f\u5728\u6811\u8393\u6d3e\u8fd9\u6837\u7684\u5d4c\u5165\u5f0f\u8bbe\u5907\u4e0a，\u4e5f\u80fd\u8fbe\u5230\u63a5\u8fd1\u5b9e\u65f6\u7684\u68c0\u6d4b\u901f\u5ea6\u3002\u5bf9\u4e8e\u5927\u591a\u6570\u76d1\u63a7\u573a\u666f（\u901a\u5e38\u9700\u898115-25\u5e27\/\u79d2），\u8fd9\u4e2a\u6027\u80fd\u5b8c\u5168\u8db3\u591f\u3002<\/p>\n
\u95ee\u98981：\u5185\u5b58\u4e0d\u8db3\u9519\u8bef<\/p>\n
RuntimeError: CUDA out of memory<\/p>\n
\u89e3\u51b3\u65b9\u6848：<\/p>\n
\u95ee\u98982：\u6a21\u578b\u4e0b\u8f7d\u5931\u8d25<\/p>\n
ConnectionError: Failed to download model<\/p>\n
\u89e3\u51b3\u65b9\u6848：<\/p>\n
\u95ee\u98983：\u7aef\u53e3\u88ab\u5360\u7528<\/p>\n
OSError: [Errno 98] Address already in use<\/p>\n
\u89e3\u51b3\u65b9\u6848：<\/p>\n
\u95ee\u9898：\u68c0\u6d4b\u7ed3\u679c\u4e0d\u51c6\u786e \u53ef\u80fd\u539f\u56e0\u548c\u89e3\u51b3\u65b9\u6848：<\/p>\n
\u8c03\u6574\u7f6e\u4fe1\u5ea6\u9608\u503c\u7684\u4ee3\u7801\u793a\u4f8b：<\/p>\n
def detect_with_threshold(image_path, confidence_threshold=0.5):
\n """\u4f7f\u7528\u81ea\u5b9a\u4e49\u7f6e\u4fe1\u5ea6\u9608\u503c\u8fdb\u884c\u68c0\u6d4b"""
\n detector = pipeline(
\n Tasks.domain_specific_object_detection,
\n model='damo\/cv_tinynas_object-detection_damoyolo_phone',
\n cache_dir='\/root\/ai-models',
\n trust_remote_code=True
\n )<\/p>\n
result = detector(image_path)<\/p>\n
# \u8fc7\u6ee4\u4f4e\u7f6e\u4fe1\u5ea6\u7684\u7ed3\u679c
\n filtered_boxes = []
\n filtered_scores = []<\/p>\n
for box, score in zip(result['boxes'], result['scores']):
\n if score >= confidence_threshold:
\n filtered_boxes.append(box)
\n filtered_scores.append(score)<\/p>\n
return {
\n 'boxes': filtered_boxes,
\n 'scores': filtered_scores,
\n 'original_count': len(result['boxes']),
\n 'filtered_count': len(filtered_boxes)
\n }<\/p>\n
\u5982\u679c\u4f60\u53d1\u73b0\u68c0\u6d4b\u901f\u5ea6\u4e0d\u591f\u5feb，\u53ef\u4ee5\u5c1d\u8bd5\u4ee5\u4e0b\u4f18\u5316：<\/p>\n
GPU\u52a0\u901f\u7684\u4ee3\u7801\u793a\u4f8b：<\/p>\n
def detect_with_gpu(image_path):
\n """\u4f7f\u7528GPU\u52a0\u901f\u68c0\u6d4b"""
\n import torch<\/p>\n
# \u68c0\u67e5GPU\u662f\u5426\u53ef\u7528
\n if not torch.cuda.is_available():
\n print("\u8b66\u544a: GPU\u4e0d\u53ef\u7528，\u5c06\u4f7f\u7528CPU")
\n device = 'cpu'
\n else:
\n device = 'cuda:0'
\n print(f"\u4f7f\u7528GPU: {torch.cuda.get_device_name(0)}")<\/p>\n
detector = pipeline(
\n Tasks.domain_specific_object_detection,
\n model='damo\/cv_tinynas_object-detection_damoyolo_phone',
\n cache_dir='\/root\/ai-models',
\n trust_remote_code=True,
\n device=device # \u6307\u5b9a\u4f7f\u7528GPU
\n )<\/p>\n
return detector(image_path)<\/p>\n
\u901a\u8fc7\u4eca\u5929\u7684\u5206\u4eab，\u4f60\u5e94\u8be5\u5df2\u7ecf\u638c\u63e1\u4e86\u5982\u4f55\u5728\u4f4e\u914d\u7f6e\u670d\u52a1\u5668\u4e0a\u90e8\u7f72\u548c\u4f7f\u7528\u8fd9\u4e2a125MB\u7684\u8f7b\u91cf\u7ea7\u624b\u673a\u68c0\u6d4b\u6a21\u578b\u3002\u6211\u4eec\u6765\u56de\u987e\u4e00\u4e0b\u5173\u952e\u70b9：<\/p>\n
\u6a21\u578b\u4f18\u52bf\u660e\u663e：<\/p>\n
\u90e8\u7f72\u4f7f\u7528\u7b80\u5355：<\/p>\n
\u9002\u7528\u573a\u666f\u5e7f\u6cdb：<\/p>\n
\u8d44\u6e90\u6d88\u8017\u6781\u4f4e：<\/p>\n
\u8fd9\u4e2a\u9879\u76ee\u7684\u4ef7\u503c\u5728\u4e8e\u5b83\u8bc1\u660e\u4e86，\u5373\u4f7f\u6ca1\u6709\u9ad8\u7aef\u786c\u4ef6，\u4e5f\u80fd\u5b9e\u73b0\u9ad8\u8d28\u91cf\u7684AI\u5e94\u7528\u3002125MB\u7684\u6a21\u578b\u5927\u5c0f，\u8ba9\u624b\u673a\u68c0\u6d4b\u8fd9\u4e2a\u529f\u80fd\u53ef\u4ee5\u90e8\u7f72\u5230\u51e0\u4e4e\u4efb\u4f55\u8bbe\u5907\u4e0a，\u5927\u5927\u964d\u4f4e\u4e86AI\u6280\u672f\u7684\u4f7f\u7528\u95e8\u69db\u3002<\/p>\n
\u5982\u679c\u4f60\u6b63\u5728\u5bfb\u627e\u4e00\u4e2a\u8f7b\u91cf\u7ea7\u3001\u9ad8\u6027\u80fd\u7684\u624b\u673a\u68c0\u6d4b\u89e3\u51b3\u65b9\u6848，\u4e0d\u59a8\u8bd5\u8bd5\u8fd9\u4e2a\u6a21\u578b\u3002\u5b83\u53ef\u80fd\u4e0d\u662f\u529f\u80fd\u6700\u5168\u7684，\u4e5f\u4e0d\u662f\u51c6\u786e\u7387\u6700\u9ad8\u7684，\u4f46\u5728\u6027\u4ef7\u6bd4\u548c\u5b9e\u7528\u6027\u65b9\u9762，\u5b83\u7edd\u5bf9\u662f\u4e00\u4e2a\u4f18\u79c0\u7684\u9009\u62e9\u3002<\/p>\n
\u83b7\u53d6\u66f4\u591aAI\u955c\u50cf<\/p>\n
\u60f3\u63a2\u7d22\u66f4\u591aAI\u955c\u50cf\u548c\u5e94\u7528\u573a\u666f？\u8bbf\u95ee CSDN\u661f\u56fe\u955c\u50cf\u5e7f\u573a，\u63d0\u4f9b\u4e30\u5bcc\u7684\u9884\u7f6e\u955c\u50cf，\u8986\u76d6\u5927\u6a21\u578b\u63a8\u7406\u3001\u56fe\u50cf\u751f\u6210\u3001\u89c6\u9891\u751f\u6210\u3001\u6a21\u578b\u5fae\u8c03\u7b49\u591a\u4e2a\u9886\u57df，\u652f\u6301\u4e00\u952e\u90e8\u7f72\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
\u5b9e\u65f6\u624b\u673a\u68c0\u6d4b-\u901a\u7528\u955c\u50cf\u90e8\u7f72：125MB\u8f7b\u91cf\u6a21\u578b\u9002\u914d\u4f4e\u914d\u670d\u52a1\u5668\u5b9e\u8df5
\n1. \u5f15\u8a00
\n\u4f60\u6709\u6ca1\u6709\u9047\u5230\u8fc7\u8fd9\u6837\u7684\u573a\u666f？\u60f3\u5728\u4f1a\u8bae\u5ba4\u3001\u56fe\u4e66\u9986\u6216\u8005\u751f\u4ea7\u7ebf\u4e0a\u90e8\u7f72\u4e00\u4e2a\u624b\u673a\u68c0\u6d4b\u7cfb\u7edf，\u7528\u6765\u7edf\u8ba1\u624b\u673a\u4f7f\u7528\u60c5\u51b5\u6216\u8005\u786e\u4fdd\u7279\u5b9a\u533a\u57df\u7684\u5b89\u5168，\u7ed3\u679c\u53d1\u73b0\u670d\u52a1\u5668\u914d\u7f6e\u592a\u4f4e，\u8dd1\u4e0d\u52a8\u90a3\u4e9b\u52a8\u8f84\u51e0\u4e2aG\u7684\u68c0\u6d4b\u6a21\u578b\u3002
\n\u6211\u4e4b\u524d\u5c31\u78b0\u5230\u8fc7\u8fd9\u4e2a\u95ee\u9898\u3002\u5ba2\u6237\u6709\u4e00\u53f0\u8001\u65e7\u7684\u670d\u52a1\u5668，\u5185\u5b58\u53ea\u67094GB，CPU\u4e5f\u662f\u597d\u51e0\u5e74\u524d\u7684\u578b\u53f7，\u4f46\u9700\u8981\u5728\u5165\u53e3\u5904<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[7591,9826,1665,523],"topic":[],"class_list":["post-86987","post","type-post","status-publish","format-standard","hentry","category-server","tag-ai","tag-9826","tag-1665","tag-523"],"yoast_head":"\n