{"id":121447,"date":"2026-11-02T20:32:49","date_gmt":"2026-11-02T12:32:49","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/121447.html"},"modified":"2026-11-02T20:32:49","modified_gmt":"2026-11-02T12:32:49","slug":"%e6%95%b0%e6%8d%ae%e9%9b%8638-%e9%9d%92%e5%85%89%e7%9c%bc%e6%a3%80%e6%b5%8b%e6%95%b0%e6%8d%ae%e9%9b%86%ef%bc%8c%e9%80%82%e5%90%88%e5%8c%bb%e5%ad%a6%e5%bd%b1%e5%83%8f%e5%88%86%e6%9e%90%e3%80%81","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/121447.html","title":{"rendered":"\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1"},"content":{"rendered":"

\u3010\u6df1\u5ea6\u5b66\u4e60\u6570\u636e\u96c6\u63a8\u8350\u3011\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6，\u5305\u542b\u773c\u5e95\u56fe\u50cf\u4e0e650\u6761\u6807\u6ce8\u8bb0\u5f55，\u9644Python\u4ee3\u7801\u53ca\u6bd5\u8bbe\u9898\u76ee<\/h2>\n

\u4eca\u5929\u5206\u4eab\u4e00\u4e2a\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1\u7684\u516c\u5f00\u6570\u636e\u96c6<\/p>\n

\u8be5\u6570\u636e\u96c6\u4e3b\u8981\u7528\u4e8e\u9752\u5149\u773c\u76f8\u5173\u5f71\u50cf\u8bc6\u522b\u7814\u7a76，\u5305\u542bACRIMA\u3001ORIGA\u7b49\u6570\u636e\u76ee\u5f55，\u4ee5\u53ca\u8bb0\u5f55\u56fe\u50cf\u540d\u79f0\u3001\u89c6\u676f\u76d8\u6bd4\u548c\u9752\u5149\u773c\u6807\u7b7e\u7684CSV\u6587\u4ef6\u3002<\/p>\n

\u5bf9\u4e8e\u8ba1\u7b97\u673a\u3001\u4eba\u5de5\u667a\u80fd\u3001\u533b\u5b66\u4fe1\u606f\u5de5\u7a0b\u3001\u751f\u7269\u533b\u5b66\u5de5\u7a0b\u7b49\u4e13\u4e1a\u7684\u540c\u5b66，\u53ef\u4ee5\u56f4\u7ed5\u9752\u5149\u773c\u56fe\u50cf\u5206\u7c7b\u3001\u8fc1\u79fb\u5b66\u4e60\u3001\u591a\u6a21\u578b\u6027\u80fd\u5bf9\u6bd4\u3001\u89c6\u676f\u76d8\u6bd4\u8f85\u52a9\u5206\u6790\u548c\u533b\u5b66\u5f71\u50cf\u8bc6\u522b\u7cfb\u7edf\u5f00\u5c55\u6bd5\u4e1a\u8bbe\u8ba1\u3002<\/p>\n

\u4e00\u3001\u6570\u636e\u96c6\u57fa\u672c\u4ecb\u7ecd<\/h3>\n

1.1 \u6570\u636e\u96c6\u6982\u51b5<\/h4>\n\n\u9879\u76ee\u5185\u5bb9<\/tr>\n\n\n\n\n\n\n\n\n\n\n\n
\u4e2d\u6587\u540d\u79f0<\/td>\n\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6<\/td>\n<\/tr>\n
\u6570\u636e\u5927\u5c0f<\/td>\n454.78MB<\/td>\n<\/tr>\n
\u9875\u9762\u6587\u4ef6\u603b\u6570<\/td>\n2,662\u4e2a<\/td>\n<\/tr>\n
\u6570\u636e\u76ee\u5f55<\/td>\nACRIMA\u3001Fundus_Train_Val_Data\u3001ORIGA<\/td>\n<\/tr>\n
\u6807\u7b7e\u6587\u4ef6<\/td>\nglaucoma.csv<\/td>\n<\/tr>\n
CSV\u8bb0\u5f55\u6570\u91cf<\/td>\n650\u6761<\/td>\n<\/tr>\n
CSV\u5b57\u6bb5\u6570\u91cf<\/td>\n5\u5217<\/td>\n<\/tr>\n
\u4e3b\u8981\u4efb\u52a1<\/td>\n\u9752\u5149\u773c\u4e8c\u5206\u7c7b<\/td>\n<\/tr>\n
\u6269\u5c55\u4efb\u52a1<\/td>\nExpCDR\u6307\u6807\u5206\u6790\u3001\u6a21\u578b\u5bf9\u6bd4<\/td>\n<\/tr>\n
\u63a8\u8350\u7b97\u6cd5<\/td>\nResNet\u3001DenseNet\u3001EfficientNet\u3001MobileNet<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

1.2 \u4ec0\u4e48\u662f\u9752\u5149\u773c？<\/h4>\n

\u9752\u5149\u773c\u662f\u4e00\u7ec4\u53ef\u80fd\u9020\u6210\u89c6\u795e\u7ecf\u635f\u4f24\u7684\u773c\u90e8\u75be\u75c5，\u662f\u4e0d\u53ef\u9006\u89c6\u529b\u635f\u4f24\u7684\u91cd\u8981\u539f\u56e0\u4e4b\u4e00\u3002<\/p>\n

\u5728\u773c\u5e95\u5f71\u50cf\u5206\u6790\u4e2d，\u89c6\u76d8\u3001\u89c6\u676f\u7b49\u7ed3\u6784\u53ef\u4ee5\u4e3a\u9752\u5149\u773c\u98ce\u9669\u8bc4\u4f30\u63d0\u4f9b\u4e00\u5b9a\u53c2\u8003\u3002<\/p>\n

\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u53ef\u4ee5\u5b66\u4e60\u773c\u5e95\u56fe\u50cf\u4e2d\u7684\u89c6\u89c9\u7279\u5f81，\u5f00\u5c55\u9752\u5149\u773c\u76f8\u5173\u5206\u7c7b\u7814\u7a76\u3002<\/p>\n

\u4f46\u9700\u8981\u5f3a\u8c03，\u56fe\u50cf\u5206\u7c7b\u7ed3\u679c\u4e0d\u80fd\u76f4\u63a5\u66ff\u4ee3\u4e34\u5e8a\u9752\u5149\u773c\u8bca\u65ad\u3002\u771f\u5b9e\u8bca\u65ad\u8fd8\u9700\u8981\u7ed3\u5408\u773c\u538b\u3001\u89c6\u91ce\u68c0\u67e5\u3001\u89c6\u795e\u7ecf\u8bc4\u4f30\u7b49\u4e34\u5e8a\u4fe1\u606f\u3002
\n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/p>\n

\u4e8c\u3001\u6570\u636e\u96c6\u5305\u542b\u54ea\u4e9b\u5185\u5bb9？<\/h3>\n

\u6839\u636e\u63d0\u4f9b\u7684Kaggle\u76ee\u5f55\u4fe1\u606f，\u4e3b\u8981\u5305\u542b\u4ee5\u4e0b\u5185\u5bb9：<\/p>\n

Detection\/
\n\u251c\u2500\u2500 ACRIMA\/
\n\u251c\u2500\u2500 Fundus_Train_Val_Data\/
\n\u251c\u2500\u2500 ORIGA\/
\n\u2514\u2500\u2500 glaucoma.csv<\/p>\n

\u4e0d\u540c\u76ee\u5f55\u53ef\u80fd\u5177\u6709\u4e0d\u540c\u7684\u56fe\u50cf\u7ec4\u7ec7\u65b9\u5f0f\u3001\u6807\u7b7e\u5f62\u5f0f\u548c\u6570\u636e\u6765\u6e90\u3002<\/p>\n

\u56e0\u6b64，\u5efa\u8bae\u5148\u5b8c\u6210\u76ee\u5f55\u68c0\u67e5，\u518d\u786e\u5b9a\u91c7\u7528\u54ea\u4e2a\u5b50\u96c6\u8fdb\u884c\u8bad\u7ec3\u3002<\/p>\n

2.1 glaucoma.csv\u5b57\u6bb5\u8bf4\u660e<\/h4>\n

CSV\u5305\u542b5\u4e2a\u5b57\u6bb5：<\/p>\n\n\u5b57\u6bb5\u542b\u4e49\u7528\u9014<\/tr>\n\n\n\n\n\n\n
Filename<\/td>\n\u56fe\u50cf\u6587\u4ef6\u540d\u79f0<\/td>\n\u5173\u8054\u56fe\u50cf<\/td>\n<\/tr>\n
ExpCDR<\/td>\n\u4e13\u5bb6\u8bc4\u4f30\u7684\u676f\u76d8\u6bd4\u76f8\u5173\u6570\u503c<\/td>\n\u8f85\u52a9\u7279\u5f81\u5206\u6790<\/td>\n<\/tr>\n
Eye<\/td>\n\u5de6\u773c\u6216\u53f3\u773c\u6807\u8bc6<\/td>\n\u6570\u636e\u7edf\u8ba1<\/td>\n<\/tr>\n
Set<\/td>\n\u6570\u636e\u96c6\u5206\u7ec4\u6807\u8bc6<\/td>\n\u6570\u636e\u5206\u7ec4\u5206\u6790<\/td>\n<\/tr>\n
Glaucoma<\/td>\n\u9752\u5149\u773c\u6807\u7b7e<\/td>\n\u5206\u7c7b\u76ee\u6807<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

\u4f8b\u5982：<\/p>\n\nFilenameExpCDREyeSetGlaucoma<\/tr>\n\n\n\n\n\n
001.jpg<\/td>\n0.7097<\/td>\nOD<\/td>\nA<\/td>\n0<\/td>\n<\/tr>\n
002.jpg<\/td>\n0.6953<\/td>\nOS<\/td>\nA<\/td>\n0<\/td>\n<\/tr>\n
036.jpg<\/td>\n0.5097<\/td>\nOD<\/td>\nA<\/td>\n1<\/td>\n<\/tr>\n
041.jpg<\/td>\n0.6131<\/td>\nOD<\/td>\nB<\/td>\n1<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

\u4ee5\u4e0a\u8bb0\u5f55\u6765\u81ea\u63d0\u4f9b\u7684\u6570\u636e\u9884\u89c8\u3002<\/p>\n

\u5176\u4e2d：<\/p>\n

    \n
  • OD：\u53f3\u773c\u3002<\/li>\n
  • OS：\u5de6\u773c\u3002<\/li>\n
  • ExpCDR：\u89c6\u676f\u76d8\u6bd4\u76f8\u5173\u6307\u6807\u3002<\/li>\n
  • Glaucoma：\u4e8c\u5206\u7c7b\u6807\u7b7e，0\u548c1\u3002<\/li>\n<\/ul>\n

    \u8fd9\u91cc\u7684Set\u5b57\u6bb5\u51fa\u73b0A\u3001B\u4e24\u79cd\u53d6\u503c，\u4f46\u4ec5\u51ed\u9884\u89c8\u4e0d\u80fd\u786e\u8ba4\u5176\u5177\u4f53\u5212\u5206\u542b\u4e49，\u9700\u8981\u67e5\u770b\u539f\u59cb\u6570\u636e\u8bf4\u660e\u3002<\/p>\n

    \u4e09\u3001\u9752\u5149\u773c\u6807\u7b7e\u5206\u5e03<\/h3>\n

    \u6839\u636eglaucoma.csv\u63d0\u4f9b\u7684\u7edf\u8ba1：<\/p>\n\n\u6807\u7b7e\u542b\u4e49\u6837\u672c\u6570\u91cf\u5360\u6bd4<\/tr>\n\n\n\n\n
    0<\/td>\n\u975e\u9752\u5149\u773c<\/td>\n482<\/td>\n74.15%<\/td>\n<\/tr>\n
    1<\/td>\n\u9752\u5149\u773c<\/td>\n168<\/td>\n25.85%<\/td>\n<\/tr>\n
    \u5408\u8ba1<\/td>\n<\/td>\n650<\/td>\n100%<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    \u53ef\u4ee5\u770b\u51fa，\u975e\u9752\u5149\u773c\u6837\u672c\u6570\u91cf\u660e\u663e\u591a\u4e8e\u9752\u5149\u773c\u6837\u672c\u3002<\/p>\n

    \u56e0\u6b64，\u8bad\u7ec3\u6a21\u578b\u65f6\u4e0d\u80fd\u53ea\u5173\u6ce8Accuracy，\u8fd8\u5e94\u91cd\u70b9\u8bc4\u4ef7\u9752\u5149\u773c\u7c7b\u522b\u7684Recall\u3001Precision\u3001F1-score\u548cROC-AUC\u3002<\/p>\n

    \u5c24\u5176\u662f\u533b\u5b66\u5f71\u50cf\u5206\u7c7b\u4efb\u52a1，\u6f0f\u68c0\u4e0e\u8bef\u68c0\u7684\u5b9e\u9645\u5f71\u54cd\u4e0d\u540c，\u9700\u8981\u7ed3\u5408\u7814\u7a76\u76ee\u6807\u5408\u7406\u9009\u62e9\u8bc4\u4ef7\u6307\u6807\u3002<\/p>\n

    \u56db\u3001ExpCDR\u6307\u6807\u6709\u4ec0\u4e48\u4f5c\u7528？<\/h3>\n

    ExpCDR\u662f\u8fd9\u4efd\u6570\u636e\u96c6\u6bd4\u8f83\u6709\u7814\u7a76\u4ef7\u503c\u7684\u5b57\u6bb5\u3002<\/p>\n

    CDR\u901a\u5e38\u6307\u676f\u76d8\u6bd4（Cup-to-Disc Ratio），\u7528\u4e8e\u63cf\u8ff0\u89c6\u676f\u4e0e\u89c6\u76d8\u4e4b\u95f4\u7684\u76f8\u5bf9\u6bd4\u4f8b\u3002<\/p>\n

    \u6570\u636e\u9884\u89c8\u4e2d\u7684ExpCDR\u53d6\u503c\u5927\u7ea6\u5206\u5e03\u57280.16\u81f30.96\u4e4b\u95f4\u3002<\/p>\n

    \u53ef\u4ee5\u56f4\u7ed5\u8be5\u6307\u6807\u5f00\u5c55\u4ee5\u4e0b\u5206\u6790：<\/p>\n

  • \u9752\u5149\u773c\u4e0e\u975e\u9752\u5149\u773c\u6837\u672c\u7684ExpCDR\u5206\u5e03\u5bf9\u6bd4\u3002<\/li>\n
  • ExpCDR\u4e0e\u9752\u5149\u773c\u6807\u7b7e\u4e4b\u95f4\u7684\u5173\u8054\u5206\u6790\u3002<\/li>\n
  • \u4e0d\u540cExpCDR\u533a\u95f4\u7684\u6837\u672c\u6570\u91cf\u7edf\u8ba1\u3002<\/li>\n
  • \u5355\u72ec\u4f7f\u7528ExpCDR\u6784\u5efa\u673a\u5668\u5b66\u4e60\u5206\u7c7b\u57fa\u7ebf\u3002<\/li>\n
  • \u6bd4\u8f83\u56fe\u50cf\u6a21\u578b\u4e0e\u56fe\u50cf\u52a0ExpCDR\u878d\u5408\u6a21\u578b\u7684\u8868\u73b0\u3002<\/li>\n

    \u9700\u8981\u6ce8\u610f：<\/p>\n

    ExpCDR\u4e0d\u80fd\u5355\u72ec\u4f5c\u4e3a\u9752\u5149\u773c\u8bca\u65ad\u7684\u7edd\u5bf9\u6807\u51c6\u3002<\/p>\n

    \u6b64\u5916，CSV\u4e2d\u5df2\u7ecf\u63d0\u4f9b\u7684\u662f\u4e13\u5bb6\u8bc4\u4f30\u6570\u503c，\u4e0d\u7b49\u4e8e\u6570\u636e\u96c6\u63d0\u4f9b\u4e86\u89c6\u76d8\u3001\u89c6\u676f\u7684\u50cf\u7d20\u7ea7\u5206\u5272\u6807\u6ce8\u3002<\/p>\n

    \u5982\u679c\u5e0c\u671b\u8bad\u7ec3U-Net\u81ea\u52a8\u5206\u5272\u89c6\u76d8\u548c\u89c6\u676f，\u9700\u8981\u989d\u5916\u786e\u8ba4\u662f\u5426\u6709\u53ef\u7528\u5206\u5272\u63a9\u819c\u3002<\/p>\n

    \u4e94\u3001\u9002\u5408\u4f7f\u7528\u54ea\u4e9b\u6df1\u5ea6\u5b66\u4e60\u7b97\u6cd5？<\/h3>\n\n\u6a21\u578b\u4e3b\u8981\u7279\u70b9\u63a8\u8350\u7528\u9014<\/tr>\n\n\n\n\n\n\n\n\n
    \u81ea\u5b9a\u4e49CNN<\/td>\n\u7ed3\u6784\u7b80\u5355<\/td>\n\u57fa\u7840\u5206\u7c7b\u6a21\u578b<\/td>\n<\/tr>\n
    ResNet18<\/td>\n\u53c2\u6570\u91cf\u9002\u4e2d<\/td>\n\u57fa\u7ebf\u6a21\u578b<\/td>\n<\/tr>\n
    ResNet50<\/td>\n\u7279\u5f81\u63d0\u53d6\u80fd\u529b\u8f83\u5f3a<\/td>\n\u6a21\u578b\u5bf9\u6bd4<\/td>\n<\/tr>\n
    DenseNet121<\/td>\n\u7279\u5f81\u590d\u7528\u5145\u5206<\/td>\n\u533b\u5b66\u5f71\u50cf\u5206\u7c7b\u5bf9\u6bd4<\/td>\n<\/tr>\n
    EfficientNet-B0<\/td>\n\u517c\u987e\u6027\u80fd\u548c\u6a21\u578b\u89c4\u6a21<\/td>\n\u4f18\u5316\u5b9e\u9a8c<\/td>\n<\/tr>\n
    MobileNetV3<\/td>\n\u8f7b\u91cf\u5316\u7f51\u7edc<\/td>\n\u90e8\u7f72\u5b9e\u9a8c<\/td>\n<\/tr>\n
    ViT<\/td>\nTransformer\u89c6\u89c9\u6a21\u578b<\/td>\n\u6269\u5c55\u7814\u7a76<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    \u672c\u79d1\u6bd5\u4e1a\u8bbe\u8ba1\u5efa\u8bae\u91c7\u7528：<\/p>\n

    ResNet18\u4f5c\u4e3a\u57fa\u7ebf，DenseNet121\u548cEfficientNet-B0\u4f5c\u4e3a\u5bf9\u6bd4\u6a21\u578b\u3002<\/p>\n

    \u5982\u679c\u5e0c\u671b\u589e\u52a0\u7814\u7a76\u5185\u5bb9，\u53ef\u4ee5\u8fdb\u4e00\u6b65\u7814\u7a76\u7c7b\u522b\u4e0d\u5e73\u8861\u5904\u7406\u3001\u6570\u636e\u589e\u5f3a\u3001\u6ce8\u610f\u529b\u673a\u5236\u4ee5\u53caGrad-CAM\u53ef\u89e3\u91ca\u6027\u5206\u6790\u3002<\/p>\n

    \u516d\u3001\u4f7f\u7528\u8be5\u6570\u636e\u96c6\u53ef\u4ee5\u505a\u54ea\u4e9b\u6bd5\u4e1a\u8bbe\u8ba1？<\/h3>\n

    \u63a8\u8350\u4ee5\u4e0b8\u4e2a\u65b9\u5411：<\/p>\n\n\u5e8f\u53f7\u6bd5\u8bbe\u9898\u76ee\u63a8\u8350\u7b97\u6cd5\u96be\u5ea6<\/tr>\n\n\n\n\n\n\n\n\n\n
    1<\/td>\n\u57fa\u4e8eResNet\u7684\u9752\u5149\u773c\u773c\u5e95\u56fe\u50cf\u8bc6\u522b\u7cfb\u7edf\u8bbe\u8ba1\u4e0e\u5b9e\u73b0<\/td>\nResNet18<\/td>\n\u2605\u2605\u2605<\/td>\n<\/tr>\n
    2<\/td>\n\u57fa\u4e8eDenseNet\u7684\u9752\u5149\u773c\u773c\u5e95\u56fe\u50cf\u5206\u7c7b\u7814\u7a76<\/td>\nDenseNet121<\/td>\n\u2605\u2605\u2605<\/td>\n<\/tr>\n
    3<\/td>\n\u57fa\u4e8e\u8fc1\u79fb\u5b66\u4e60\u7684\u9752\u5149\u773c\u56fe\u50cf\u5206\u7c7b\u6a21\u578b\u5bf9\u6bd4\u7814\u7a76<\/td>\nResNet\u3001EfficientNet<\/td>\n\u2605\u2605\u2605<\/td>\n<\/tr>\n
    4<\/td>\n\u57fa\u4e8e\u6ce8\u610f\u529b\u673a\u5236\u7684\u9752\u5149\u773c\u773c\u5e95\u56fe\u50cf\u8bc6\u522b\u7814\u7a76<\/td>\nResNet+CBAM<\/td>\n\u2605\u2605\u2605\u2605<\/td>\n<\/tr>\n
    5<\/td>\n\u57fa\u4e8eMobileNetV3\u7684\u8f7b\u91cf\u5316\u9752\u5149\u773c\u56fe\u50cf\u5206\u7c7b\u7814\u7a76<\/td>\nMobileNetV3<\/td>\n\u2605\u2605\u2605\u2605<\/td>\n<\/tr>\n
    6<\/td>\n\u57fa\u4e8eGrad-CAM\u7684\u9752\u5149\u773c\u5206\u7c7b\u6a21\u578b\u53ef\u89e3\u91ca\u6027\u7814\u7a76<\/td>\nResNet+Grad-CAM<\/td>\n\u2605\u2605\u2605\u2605<\/td>\n<\/tr>\n
    7<\/td>\n\u878d\u5408\u89c6\u676f\u76d8\u6bd4\u7279\u5f81\u7684\u9752\u5149\u773c\u8f85\u52a9\u5206\u7c7b\u65b9\u6cd5\u7814\u7a76<\/td>\nCNN+MLP<\/td>\n\u2605\u2605\u2605\u2605<\/td>\n<\/tr>\n
    8<\/td>\n\u57fa\u4e8e\u6df1\u5ea6\u5b66\u4e60\u7684\u9752\u5149\u773c\u773c\u5e95\u56fe\u50cf\u5206\u6790\u4e0e\u53ef\u89c6\u5316\u7cfb\u7edf<\/td>\nPyTorch+Flask+Vue<\/td>\n\u2605\u2605\u2605<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    \u63a8\u8350\u9898\u76ee1：\u57fa\u4e8eResNet\u7684\u9752\u5149\u773c\u773c\u5e95\u56fe\u50cf\u8bc6\u522b\u7cfb\u7edf\u8bbe\u8ba1\u4e0e\u5b9e\u73b0<\/h4>\n

    \u63a8\u8350\u6307\u6570：\u2605\u2605\u2605\u2605\u2605<\/p>\n

    \u9002\u5408\u672c\u79d1\u8ba1\u7b97\u673a\u3001\u8f6f\u4ef6\u5de5\u7a0b\u548c\u4eba\u5de5\u667a\u80fd\u4e13\u4e1a\u3002<\/p>\n

    \u4e3b\u8981\u7814\u7a76\u5185\u5bb9：<\/p>\n

  • \u6570\u636e\u96c6\u6574\u7406\u4e0e\u56fe\u50cf\u6807\u7b7e\u5339\u914d\u3002<\/li>\n
  • \u773c\u5e95\u56fe\u50cf\u9884\u5904\u7406\u3002<\/li>\n
  • ResNet18\u8fc1\u79fb\u5b66\u4e60\u3002<\/li>\n
  • ResNet50\u6a21\u578b\u5bf9\u6bd4\u3002<\/li>\n
  • \u9752\u5149\u773c\u4e8c\u5206\u7c7b\u6027\u80fd\u8bc4\u4ef7\u3002<\/li>\n
  • Grad-CAM\u7ed3\u679c\u53ef\u89c6\u5316\u3002<\/li>\n
  • Web\u56fe\u50cf\u8bc6\u522b\u7cfb\u7edf\u5f00\u53d1\u3002<\/li>\n

    \u6700\u7ec8\u5b9e\u73b0\u4e0a\u4f20\u773c\u5e95\u56fe\u7247，\u7cfb\u7edf\u8f93\u51fa\u6a21\u578b\u9884\u6d4b\u7c7b\u522b\u3001\u5206\u7c7b\u7f6e\u4fe1\u5ea6\u53ca\u53ef\u89c6\u5316\u7ed3\u679c\u3002<\/p>\n

    \u63a8\u8350\u9898\u76ee2：\u57fa\u4e8e\u6ce8\u610f\u529b\u673a\u5236\u7684\u9752\u5149\u773c\u773c\u5e95\u56fe\u50cf\u5206\u7c7b\u7814\u7a76<\/h4>\n

    \u63a8\u8350\u6307\u6570：\u2605\u2605\u2605\u2605\u2606<\/p>\n

    \u5982\u679c\u5e0c\u671b\u589e\u52a0\u7b97\u6cd5\u7814\u7a76\u5185\u5bb9，\u53ef\u4ee5\u8003\u8651\u8fd9\u4e2a\u65b9\u5411\u3002<\/p>\n

    \u7814\u7a76\u76ee\u6807：<\/p>\n

    \u5728ResNet\u6a21\u578b\u4e2d\u52a0\u5165\u6ce8\u610f\u529b\u6a21\u5757，\u7814\u7a76\u4e0d\u540c\u7279\u5f81\u63d0\u53d6\u65b9\u5f0f\u5bf9\u9752\u5149\u773c\u56fe\u50cf\u5206\u7c7b\u6548\u679c\u7684\u5f71\u54cd\u3002<\/p>\n

    \u5b9e\u9a8c\u8bbe\u8ba1：<\/p>\n\n\u5b9e\u9a8c\u6a21\u578b<\/tr>\n\n\n\n\n\n
    Baseline<\/td>\nResNet18<\/td>\n<\/tr>\n
    \u5b9e\u9a8c1<\/td>\nResNet18+SE<\/td>\n<\/tr>\n
    \u5b9e\u9a8c2<\/td>\nResNet18+CBAM<\/td>\n<\/tr>\n
    \u5b9e\u9a8c3<\/td>\nResNet18+ECA<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    \u91cd\u70b9\u6bd4\u8f83Accuracy\u3001Recall\u3001F1-score\u548cROC-AUC\u3002<\/p>\n

    \u63a8\u8350\u9898\u76ee3：\u878d\u5408\u89c6\u676f\u76d8\u6bd4\u7279\u5f81\u7684\u9752\u5149\u773c\u8f85\u52a9\u5206\u7c7b\u65b9\u6cd5\u7814\u7a76<\/h4>\n

    \u63a8\u8350\u6307\u6570：\u2605\u2605\u2605\u2605\u2605<\/p>\n

    \u8fd9\u4e2a\u65b9\u5411\u53ef\u4ee5\u5229\u7528\u6570\u636e\u96c6\u4e2d\u63d0\u4f9b\u7684ExpCDR\u5b57\u6bb5，\u533a\u522b\u4e8e\u666e\u901a\u7684\u56fe\u50cf\u5206\u7c7b\u9879\u76ee\u3002<\/p>\n

    \u7814\u7a76\u76ee\u6807：<\/p>\n

    \u7ed3\u5408\u773c\u5e95\u56fe\u50cf\u7279\u5f81\u548cExpCDR\u6570\u503c，\u6784\u5efa\u591a\u6a21\u6001\u7279\u5f81\u878d\u5408\u5206\u7c7b\u6a21\u578b\u3002<\/p>\n

    \u6280\u672f\u6d41\u7a0b：<\/p>\n

    \u773c\u5e95\u56fe\u50cf \u2500\u2500\u2192 ResNet \u2500\u2500\u2192 \u56fe\u50cf\u7279\u5f81 \u2500\u2500\u2510
    \n \u251c\u2500\u2500\u2192 \u7279\u5f81\u878d\u5408 \u2500\u2500\u2192 \u4e8c\u5206\u7c7b
    \nExpCDR\u6570\u503c \u2500\u2500\u2192 MLP \u2500\u2500\u2192 \u6570\u503c\u7279\u5f81 \u2500\u2500\u2500\u2518<\/p>\n

    \u63a8\u8350\u5bf9\u6bd4\u5b9e\u9a8c：<\/p>\n\n\u5b9e\u9a8c\u6a21\u578b\u65b9\u6848<\/tr>\n\n\n\n\n\n
    \u5b9e\u9a8c1<\/td>\n\u4ec5\u4f7f\u7528ExpCDR+\u903b\u8f91\u56de\u5f52<\/td>\n<\/tr>\n
    \u5b9e\u9a8c2<\/td>\n\u4ec5\u4f7f\u7528\u773c\u5e95\u56fe\u50cf+ResNet<\/td>\n<\/tr>\n
    \u5b9e\u9a8c3<\/td>\n\u56fe\u50cf\u7279\u5f81+ExpCDR\u878d\u5408<\/td>\n<\/tr>\n
    \u5b9e\u9a8c4<\/td>\n\u56fe\u50cf\u7279\u5f81+ExpCDR+\u6ce8\u610f\u529b\u6a21\u5757<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    \u9700\u8981\u6ce8\u610f，\u878d\u5408\u6a21\u578b\u4f9d\u8d56ExpCDR\u4f5c\u4e3a\u989d\u5916\u8f93\u5165\u3002\u5982\u679c\u5b9e\u9645\u7cfb\u7edf\u6ca1\u6709\u4e13\u5bb6\u63d0\u4f9b\u7684ExpCDR\u6570\u503c，\u5c31\u4e0d\u80fd\u76f4\u63a5\u4f7f\u7528\u8fd9\u4e00\u878d\u5408\u6a21\u578b\u5b8c\u6210\u9884\u6d4b，\u9664\u975e\u53e6\u884c\u5f00\u53d1\u53ef\u9760\u7684\u676f\u76d8\u6bd4\u4f30\u8ba1\u65b9\u6cd5\u3002<\/p>\n

    \u4e03\u3001Python\u4ee3\u7801：\u5206\u6790CSV\u6807\u7b7e\u5206\u5e03<\/h3>\n

    \u9996\u5148\u5bf9glaucoma.csv\u8fdb\u884c\u57fa\u7840\u7edf\u8ba1\u3002<\/p>\n

    7.1 \u5b89\u88c5\u4f9d\u8d56<\/h4>\n

    pip install<\/span> pandas numpy matplotlib seaborn scikit-learn<\/p>\n

    7.2 \u8bfb\u53d6\u6570\u636e<\/h4>\n

    import<\/span> pandas as<\/span> pd<\/p>\n

    df =<\/span> pd.<\/span>read_csv(<\/span>"glaucoma.csv"<\/span>)<\/span><\/p>\n

    print<\/span>(<\/span>"\u6570\u636e\u5f62\u72b6："<\/span>,<\/span> df.<\/span>shape)<\/span>
    \nprint<\/span>(<\/span>df.<\/span>head(<\/span>)<\/span>)<\/span><\/p>\n

    print<\/span>(<\/span>"\\\\n\u7f3a\u5931\u503c："<\/span>)<\/span>
    \nprint<\/span>(<\/span>df.<\/span>isnull(<\/span>)<\/span>.<\/span>sum<\/span>(<\/span>)<\/span>)<\/span><\/p>\n

    print<\/span>(<\/span>"\\\\n\u6807\u7b7e\u5206\u5e03："<\/span>)<\/span>
    \nprint<\/span>(<\/span>df[<\/span>"Glaucoma"<\/span>]<\/span>.<\/span>value_counts(<\/span>)<\/span>)<\/span><\/p>\n

    print<\/span>(<\/span>"\\\\nExpCDR\u7edf\u8ba1："<\/span>)<\/span>
    \nprint<\/span>(<\/span>df[<\/span>"ExpCDR"<\/span>]<\/span>.<\/span>describe(<\/span>)<\/span>)<\/span><\/p>\n

    7.3 \u53ef\u89c6\u5316\u9752\u5149\u773c\u6807\u7b7e\u5206\u5e03<\/h4>\n

    import<\/span> matplotlib.<\/span>pyplot as<\/span> plt<\/p>\n

    counts =<\/span> df[<\/span>"Glaucoma"<\/span>]<\/span>.<\/span>value_counts(<\/span>)<\/span>.<\/span>sort_index(<\/span>)<\/span><\/p>\n

    plt.<\/span>figure(<\/span>figsize=<\/span>(<\/span>6<\/span>,<\/span> 4<\/span>)<\/span>)<\/span><\/p>\n

    plt.<\/span>bar(<\/span>
    \n [<\/span>"Non-Glaucoma"<\/span>,<\/span> "Glaucoma"<\/span>]<\/span>,<\/span>
    \n [<\/span>counts.<\/span>get(<\/span>0<\/span>,<\/span> 0<\/span>)<\/span>,<\/span> counts.<\/span>get(<\/span>1<\/span>,<\/span> 0<\/span>)<\/span>]<\/span>
    \n)<\/span><\/p>\n

    plt.<\/span>xlabel(<\/span>"Class"<\/span>)<\/span>
    \nplt.<\/span>ylabel(<\/span>"Count"<\/span>)<\/span>
    \nplt.<\/span>title(<\/span>"Glaucoma Class Distribution"<\/span>)<\/span><\/p>\n

    plt.<\/span>tight_layout(<\/span>)<\/span>
    \nplt.<\/span>show(<\/span>)<\/span><\/p>\n

    7.4 ExpCDR\u4e0e\u9752\u5149\u773c\u6807\u7b7e\u5206\u6790<\/h4>\n

    import<\/span> seaborn as<\/span> sns
    \nimport<\/span> matplotlib.<\/span>pyplot as<\/span> plt<\/p>\n

    sns.<\/span>boxplot(<\/span>
    \n data=<\/span>df,<\/span>
    \n x=<\/span>"Glaucoma"<\/span>,<\/span>
    \n y=<\/span>"ExpCDR"<\/span>
    \n)<\/span><\/p>\n

    plt.<\/span>title(<\/span>"ExpCDR by Glaucoma Label"<\/span>)<\/span>
    \nplt.<\/span>xlabel(<\/span>"Glaucoma"<\/span>)<\/span>
    \nplt.<\/span>ylabel(<\/span>"ExpCDR"<\/span>)<\/span><\/p>\n

    plt.<\/span>show(<\/span>)<\/span><\/p>\n

    \u901a\u8fc7\u7bb1\u7ebf\u56fe\u53ef\u4ee5\u89c2\u5bdf\u4e0d\u540c\u7c7b\u522b\u7684ExpCDR\u5206\u5e03\u5dee\u5f02\u3002<\/p>\n

    \u4f46\u8fd9\u79cd\u5dee\u5f02\u53ea\u80fd\u8bf4\u660e\u6570\u636e\u4e2d\u7684\u7edf\u8ba1\u5173\u8054，\u4e0d\u80fd\u76f4\u63a5\u8bc1\u660e\u56e0\u679c\u5173\u7cfb\u6216\u4e34\u5e8a\u8bca\u65ad\u4ef7\u503c\u3002<\/p>\n

    \u516b\u3001Python\u4ee3\u7801：\u4f7f\u7528ExpCDR\u5efa\u7acb\u673a\u5668\u5b66\u4e60\u5206\u7c7b\u57fa\u7ebf<\/h3>\n

    \u9664\u4e86\u56fe\u50cf\u5206\u7c7b，\u8fd8\u53ef\u4ee5\u5148\u5229\u7528CSV\u4e2d\u7684\u6570\u503c\u7279\u5f81\u6784\u5efa\u4e00\u4e2a\u7b80\u5355\u7684\u673a\u5668\u5b66\u4e60\u6a21\u578b\u3002<\/p>\n

    \u8fd9\u91cc\u4f7f\u7528Logistic Regression\u3002<\/p>\n

    import<\/span> pandas as<\/span> pd<\/p>\n

    from<\/span> sklearn.<\/span>model_selection import<\/span> train_test_split
    \nfrom<\/span> sklearn.<\/span>preprocessing import<\/span> StandardScaler
    \nfrom<\/span> sklearn.<\/span>linear_model import<\/span> LogisticRegression
    \nfrom<\/span> sklearn.<\/span>pipeline import<\/span> Pipeline
    \nfrom<\/span> sklearn.<\/span>metrics import<\/span> classification_report,<\/span> roc_auc_score<\/p>\n

    df =<\/span> pd.<\/span>read_csv(<\/span>"glaucoma.csv"<\/span>)<\/span><\/p>\n

    df =<\/span> df.<\/span>dropna(<\/span>
    \n subset=<\/span>[<\/span>"ExpCDR"<\/span>,<\/span> "Glaucoma"<\/span>]<\/span>
    \n)<\/span><\/p>\n

    X =<\/span> df[<\/span>[<\/span>"ExpCDR"<\/span>]<\/span>]<\/span>
    \ny =<\/span> df[<\/span>"Glaucoma"<\/span>]<\/span>.<\/span>astype(<\/span>int<\/span>)<\/span><\/p>\n

    X_train,<\/span> X_test,<\/span> y_train,<\/span> y_test =<\/span> train_test_split(<\/span>
    \n X,<\/span>
    \n y,<\/span>
    \n test_size=<\/span>0.2<\/span>,<\/span>
    \n random_state=<\/span>42<\/span>,<\/span>
    \n stratify=<\/span>y
    \n)<\/span><\/p>\n

    model =<\/span> Pipeline(<\/span>[<\/span>
    \n (<\/span>"scaler"<\/span>,<\/span> StandardScaler(<\/span>)<\/span>)<\/span>,<\/span>
    \n (<\/span>"classifier"<\/span>,<\/span> LogisticRegression(<\/span>
    \n max_iter=<\/span>1000<\/span>,<\/span>
    \n class_weight=<\/span>"balanced"<\/span>
    \n )<\/span>)<\/span>
    \n]<\/span>)<\/span><\/p>\n

    model.<\/span>fit(<\/span>X_train,<\/span> y_train)<\/span><\/p>\n

    y_pred =<\/span> model.<\/span>predict(<\/span>X_test)<\/span>
    \ny_prob =<\/span> model.<\/span>predict_proba(<\/span>X_test)<\/span>[<\/span>:<\/span>,<\/span> 1<\/span>]<\/span><\/p>\n

    print<\/span>(<\/span>classification_report(<\/span>
    \n y_test,<\/span>
    \n y_pred,<\/span>
    \n digits=<\/span>4<\/span>,<\/span>
    \n zero_division=<\/span>0<\/span>
    \n)<\/span>)<\/span><\/p>\n

    print<\/span>(<\/span>"ROC-AUC:"<\/span>,<\/span> roc_auc_score(<\/span>y_test,<\/span> y_prob)<\/span>)<\/span><\/p>\n

    \u8fd9\u4e2a\u5b9e\u9a8c\u53ef\u4ee5\u4f5c\u4e3a\u540e\u7eed\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u6a21\u578b\u7684\u8f85\u52a9\u5bf9\u7167\u3002<\/p>\n

    \u4f46\u9700\u8981\u6ce8\u610f，\u5982\u679c\u540c\u4e00\u53d7\u8bd5\u8005\u5b58\u5728\u53cc\u773c\u6570\u636e，\u5e94\u5c3d\u53ef\u80fd\u6309\u60a3\u8005\u5206\u7ec4\u5212\u5206，\u800c\u4e0d\u662f\u76f4\u63a5\u968f\u673a\u5212\u5206\u8bb0\u5f55\u3002\u5f53\u524dCSV\u9884\u89c8\u672a\u63d0\u4f9b\u660e\u786e\u7684\u60a3\u8005ID，\u6b63\u5f0f\u7814\u7a76\u9700\u8981\u6838\u5bf9\u6570\u636e\u6765\u6e90\u548c\u5173\u8054\u4fe1\u606f\u3002<\/p>\n

    \u4e5d\u3001Python\u4ee3\u7801：ResNet18\u9752\u5149\u773c\u56fe\u50cf\u5206\u7c7b\u8bad\u7ec3<\/h3>\n

    \u5bf9\u4e8e\u56fe\u50cf\u5206\u7c7b，\u53ef\u4ee5\u4f7f\u7528PyTorch\u4e2d\u7684ResNet18\u3002<\/p>\n

    \u4f46\u7531\u4e8e\u6570\u636e\u96c6\u5305\u542b\u591a\u4e2a\u76ee\u5f55，\u9996\u5148\u9700\u8981\u5c06\u56fe\u7247\u4e0e\u5bf9\u5e94\u6807\u7b7e\u5339\u914d\u3002<\/p>\n

    \u4ee5\u4e0b\u793a\u4f8b\u5047\u8bbe\u5df2\u7ecf\u6839\u636e\u539f\u59cb\u6807\u7b7e\u6574\u7406\u4e3a：<\/p>\n

    Project\/
    \n\u251c\u2500\u2500 dataset\/
    \n\u2502 \u251c\u2500\u2500 train\/
    \n\u2502 \u2502 \u251c\u2500\u2500 normal\/
    \n\u2502 \u2502 \u2514\u2500\u2500 glaucoma\/
    \n\u2502 \u251c\u2500\u2500 val\/
    \n\u2502 \u2502 \u251c\u2500\u2500 normal\/
    \n\u2502 \u2502 \u2514\u2500\u2500 glaucoma\/
    \n\u2502 \u2514\u2500\u2500 test\/
    \n\u2502 \u251c\u2500\u2500 normal\/
    \n\u2502 \u2514\u2500\u2500 glaucoma\/
    \n\u2514\u2500\u2500 train.py<\/p>\n

    \u8fd9\u91cc\u7684normal\u548cglaucoma\u662f\u6574\u7406\u540e\u7684\u793a\u4f8b\u76ee\u5f55\u540d\u79f0，\u5e76\u975e\u65ad\u8a00\u539f\u59cbKaggle\u6570\u636e\u5c31\u662f\u8fd9\u4e00\u7ed3\u6784\u3002<\/p>\n

    9.1 \u5b89\u88c5\u4f9d\u8d56<\/h4>\n

    pip install<\/span> torch torchvision scikit-learn pillow<\/p>\n

    9.2 ResNet18\u8bad\u7ec3\u4ee3\u7801<\/h4>\n

    import<\/span> copy
    \nimport<\/span> torch
    \nimport<\/span> torch.<\/span>nn as<\/span> nn<\/p>\n

    from<\/span> torchvision import<\/span> datasets,<\/span> transforms,<\/span> models
    \nfrom<\/span> torch.<\/span>utils.<\/span>data import<\/span> DataLoader
    \nfrom<\/span> sklearn.<\/span>metrics import<\/span> classification_report<\/p>\n

    DATA_DIR =<\/span> "dataset"<\/span>
    \nEPOCHS =<\/span> 20<\/span>
    \nBATCH_SIZE =<\/span> 16<\/span>
    \nLR =<\/span> 1e-4<\/span><\/p>\n

    device =<\/span> torch.<\/span>device(<\/span>
    \n "cuda"<\/span> if<\/span> torch.<\/span>cuda.<\/span>is_available(<\/span>)<\/span> else<\/span> "cpu"<\/span>
    \n)<\/span><\/p>\n

    train_transform =<\/span> transforms.<\/span>Compose(<\/span>[<\/span>
    \n transforms.<\/span>Resize(<\/span>(<\/span>224<\/span>,<\/span> 224<\/span>)<\/span>)<\/span>,<\/span>
    \n transforms.<\/span>RandomHorizontalFlip(<\/span>)<\/span>,<\/span>
    \n transforms.<\/span>RandomRotation(<\/span>10<\/span>)<\/span>,<\/span>
    \n transforms.<\/span>ToTensor(<\/span>)<\/span>,<\/span>
    \n transforms.<\/span>Normalize(<\/span>
    \n [<\/span>0.485<\/span>,<\/span> 0.456<\/span>,<\/span> 0.406<\/span>]<\/span>,<\/span>
    \n [<\/span>0.229<\/span>,<\/span> 0.224<\/span>,<\/span> 0.225<\/span>]<\/span>
    \n )<\/span>
    \n]<\/span>)<\/span><\/p>\n

    eval_transform =<\/span> transforms.<\/span>Compose(<\/span>[<\/span>
    \n transforms.<\/span>Resize(<\/span>(<\/span>224<\/span>,<\/span> 224<\/span>)<\/span>)<\/span>,<\/span>
    \n transforms.<\/span>ToTensor(<\/span>)<\/span>,<\/span>
    \n transforms.<\/span>Normalize(<\/span>
    \n [<\/span>0.485<\/span>,<\/span> 0.456<\/span>,<\/span> 0.406<\/span>]<\/span>,<\/span>
    \n [<\/span>0.229<\/span>,<\/span> 0.224<\/span>,<\/span> 0.225<\/span>]<\/span>
    \n )<\/span>
    \n]<\/span>)<\/span><\/p>\n

    train_dataset =<\/span> datasets.<\/span>ImageFolder(<\/span>
    \n f"<\/span>{<\/span>DATA_DIR}<\/span><\/span>\/train"<\/span><\/span>,<\/span>
    \n transform=<\/span>train_transform
    \n)<\/span><\/p>\n

    val_dataset =<\/span> datasets.<\/span>ImageFolder(<\/span>
    \n f"<\/span>{<\/span>DATA_DIR}<\/span><\/span>\/val"<\/span><\/span>,<\/span>
    \n transform=<\/span>eval_transform
    \n)<\/span><\/p>\n

    test_dataset =<\/span> datasets.<\/span>ImageFolder(<\/span>
    \n f"<\/span>{<\/span>DATA_DIR}<\/span><\/span>\/test"<\/span><\/span>,<\/span>
    \n transform=<\/span>eval_transform
    \n)<\/span><\/p>\n

    assert<\/span> train_dataset.<\/span>classes ==<\/span> val_dataset.<\/span>classes
    \nassert<\/span> train_dataset.<\/span>classes ==<\/span> test_dataset.<\/span>classes<\/p>\n

    classes =<\/span> train_dataset.<\/span>classes<\/p>\n

    train_loader =<\/span> DataLoader(<\/span>
    \n train_dataset,<\/span>
    \n batch_size=<\/span>BATCH_SIZE,<\/span>
    \n shuffle=<\/span>True<\/span>
    \n)<\/span><\/p>\n

    val_loader =<\/span> DataLoader(<\/span>
    \n val_dataset,<\/span>
    \n batch_size=<\/span>BATCH_SIZE,<\/span>
    \n shuffle=<\/span>False<\/span>
    \n)<\/span><\/p>\n

    test_loader =<\/span> DataLoader(<\/span>
    \n test_dataset,<\/span>
    \n batch_size=<\/span>BATCH_SIZE,<\/span>
    \n shuffle=<\/span>False<\/span>
    \n)<\/span><\/p>\n

    model =<\/span> models.<\/span>resnet18(<\/span>
    \n weights=<\/span>models.<\/span>ResNet18_Weights.<\/span>DEFAULT
    \n)<\/span><\/p>\n

    model.<\/span>fc =<\/span> nn.<\/span>Linear(<\/span>
    \n model.<\/span>fc.<\/span>in_features,<\/span>
    \n 2<\/span>
    \n)<\/span><\/p>\n

    model =<\/span> model.<\/span>to(<\/span>device)<\/span><\/p>\n

    criterion =<\/span> nn.<\/span>CrossEntropyLoss(<\/span>)<\/span><\/p>\n

    optimizer =<\/span> torch.<\/span>optim.<\/span>Adam(<\/span>
    \n model.<\/span>parameters(<\/span>)<\/span>,<\/span>
    \n lr=<\/span>LR
    \n)<\/span><\/p>\n

    best_acc =<\/span> –<\/span>1<\/span>
    \nbest_state =<\/span> None<\/span><\/p>\n

    for<\/span> epoch in<\/span> range<\/span>(<\/span>EPOCHS)<\/span>:<\/span><\/p>\n

    model.<\/span>train(<\/span>)<\/span><\/p>\n

    train_loss =<\/span> 0<\/span><\/p>\n

    for<\/span> images,<\/span> labels in<\/span> train_loader:<\/span><\/p>\n

    images =<\/span> images.<\/span>to(<\/span>device)<\/span>
    \n labels =<\/span> labels.<\/span>to(<\/span>device)<\/span><\/p>\n

    optimizer.<\/span>zero_grad(<\/span>)<\/span><\/p>\n

    outputs =<\/span> model(<\/span>images)<\/span>
    \n loss =<\/span> criterion(<\/span>outputs,<\/span> labels)<\/span><\/p>\n

    loss.<\/span>backward(<\/span>)<\/span>
    \n optimizer.<\/span>step(<\/span>)<\/span><\/p>\n

    train_loss +=<\/span> loss.<\/span>item(<\/span>)<\/span><\/p>\n

    model.<\/span>eval<\/span>(<\/span>)<\/span><\/p>\n

    correct =<\/span> 0<\/span>
    \n total =<\/span> 0<\/span><\/p>\n

    with<\/span> torch.<\/span>no_grad(<\/span>)<\/span>:<\/span>
    \n for<\/span> images,<\/span> labels in<\/span> val_loader:<\/span><\/p>\n

    images =<\/span> images.<\/span>to(<\/span>device)<\/span>
    \n labels =<\/span> labels.<\/span>to(<\/span>device)<\/span><\/p>\n

    outputs =<\/span> model(<\/span>images)<\/span>
    \n preds =<\/span> outputs.<\/span>argmax(<\/span>dim=<\/span>1<\/span>)<\/span><\/p>\n

    correct +=<\/span> (<\/span>preds ==<\/span> labels)<\/span>.<\/span>sum<\/span>(<\/span>)<\/span>.<\/span>item(<\/span>)<\/span>
    \n total +=<\/span> labels.<\/span>size(<\/span>0<\/span>)<\/span><\/p>\n

    val_acc =<\/span> correct \/<\/span> total<\/p>\n

    print<\/span>(<\/span>
    \n f"Epoch <\/span>{<\/span>epoch+<\/span>1<\/span>}<\/span><\/span>\/<\/span>{<\/span>EPOCHS}<\/span><\/span> "<\/span><\/span>
    \n f"Loss=<\/span>{<\/span>train_loss\/<\/span>len<\/span>(<\/span>train_loader)<\/span>:<\/span>.4f<\/span>}<\/span><\/span> "<\/span><\/span>
    \n f"Val Acc=<\/span>{<\/span>val_acc:<\/span>.4f<\/span>}<\/span><\/span>"<\/span><\/span>
    \n )<\/span><\/p>\n

    if<\/span> val_acc ><\/span> best_acc:<\/span>
    \n best_acc =<\/span> val_acc
    \n best_state =<\/span> copy.<\/span>deepcopy(<\/span>
    \n model.<\/span>state_dict(<\/span>)<\/span>
    \n )<\/span><\/p>\n

    torch.<\/span>save(<\/span>
    \n {<\/span>
    \n "model_state_dict"<\/span>:<\/span> best_state,<\/span>
    \n "classes"<\/span>:<\/span> classes
    \n }<\/span>,<\/span>
    \n "best_resnet18.pth"<\/span>
    \n)<\/span><\/p>\n

    model.<\/span>load_state_dict(<\/span>best_state)<\/span>
    \nmodel.<\/span>eval<\/span>(<\/span>)<\/span><\/p>\n

    y_true =<\/span> [<\/span>]<\/span>
    \ny_pred =<\/span> [<\/span>]<\/span><\/p>\n

    with<\/span> torch.<\/span>no_grad(<\/span>)<\/span>:<\/span>
    \n for<\/span> images,<\/span> labels in<\/span> test_loader:<\/span><\/p>\n

    images =<\/span> images.<\/span>to(<\/span>device)<\/span><\/p>\n

    outputs =<\/span> model(<\/span>images)<\/span>
    \n preds =<\/span> outputs.<\/span>argmax(<\/span>dim=<\/span>1<\/span>)<\/span><\/p>\n

    y_true.<\/span>extend(<\/span>labels.<\/span>numpy(<\/span>)<\/span>.<\/span>tolist(<\/span>)<\/span>)<\/span>
    \n y_pred.<\/span>extend(<\/span>preds.<\/span>cpu(<\/span>)<\/span>.<\/span>numpy(<\/span>)<\/span>.<\/span>tolist(<\/span>)<\/span>)<\/span><\/p>\n

    print<\/span>(<\/span>classification_report(<\/span>
    \n y_true,<\/span>
    \n y_pred,<\/span>
    \n target_names=<\/span>classes,<\/span>
    \n digits=<\/span>4<\/span>,<\/span>
    \n zero_division=<\/span>0<\/span>
    \n)<\/span>)<\/span><\/p>\n

    \u4e0a\u8ff0\u4ee3\u7801\u7528\u4e8e\u5b8c\u6210\u57fa\u7840\u56fe\u50cf\u4e8c\u5206\u7c7b\u5b9e\u9a8c\u3002<\/p>\n

    \u6b63\u5f0f\u5b9e\u9a8c\u5e94\u91cd\u70b9\u4fdd\u8bc1\u56fe\u50cf\u6807\u7b7e\u6b63\u786e\u3001\u60a3\u8005\u7ea7\u6570\u636e\u9694\u79bb，\u4ee5\u53ca\u8bad\u7ec3\u96c6\u3001\u9a8c\u8bc1\u96c6\u548c\u6d4b\u8bd5\u96c6\u4e4b\u95f4\u4e0d\u5b58\u5728\u91cd\u590d\u6837\u672c\u3002
    \n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
    \n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/p>\n

    \u5341\u3001\u6bd5\u4e1a\u8bbe\u8ba1\u53ef\u4ee5\u589e\u52a0\u54ea\u4e9b\u4f18\u5316\u5b9e\u9a8c？<\/h3>\n

    10.1 \u591a\u6a21\u578b\u5bf9\u6bd4<\/h4>\n\n\u6a21\u578bAccuracyRecallF1-scoreROC-AUC<\/tr>\n\n\n\n\n\n\n
    ResNet18<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n<\/tr>\n
    ResNet50<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n<\/tr>\n
    DenseNet121<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n<\/tr>\n
    EfficientNet-B0<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n<\/tr>\n
    MobileNetV3<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n\u5f85\u5b9e\u9a8c<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

    10.2 \u7c7b\u522b\u4e0d\u5e73\u8861\u4f18\u5316<\/h4>\n

    \u9488\u5bf9CSV\u4e2d\u9752\u5149\u773c\u6837\u672c\u8f83\u5c11\u7684\u95ee\u9898，\u53ef\u4ee5\u6bd4\u8f83：<\/p>\n

      \n
    • \u666e\u901a\u4ea4\u53c9\u71b5\u3002<\/li>\n
    • \u52a0\u6743\u4ea4\u53c9\u71b5\u3002<\/li>\n
    • Focal Loss\u3002<\/li>\n
    • WeightedRandomSampler\u3002<\/li>\n<\/ul>\n

      10.3 \u6570\u636e\u589e\u5f3a\u5b9e\u9a8c<\/h4>\n

      \u53ef\u4ee5\u6bd4\u8f83：<\/p>\n

        \n
      • \u57fa\u7840\u9884\u5904\u7406\u3002<\/li>\n
      • \u968f\u673a\u7ffb\u8f6c\u3002<\/li>\n
      • \u5c0f\u89d2\u5ea6\u65cb\u8f6c\u3002<\/li>\n
      • \u4eae\u5ea6\u548c\u5bf9\u6bd4\u5ea6\u8c03\u6574\u3002<\/li>\n
      • \u4e0d\u540c\u589e\u5f3a\u7ec4\u5408\u3002<\/li>\n<\/ul>\n

        \u533b\u5b66\u5f71\u50cf\u589e\u5f3a\u5e94\u5c3d\u91cf\u4fdd\u7559\u771f\u5b9e\u89e3\u5256\u7ed3\u6784，\u907f\u514d\u4e0d\u5408\u7406\u53d8\u6362\u5f71\u54cd\u7814\u7a76\u7ed3\u679c\u3002<\/p>\n

        10.4 Grad-CAM\u53ef\u89e3\u91ca\u6027\u5206\u6790<\/h4>\n

        \u901a\u8fc7Grad-CAM\u751f\u6210\u6a21\u578b\u5173\u6ce8\u533a\u57df\u70ed\u529b\u56fe，\u5206\u6790\u6a21\u578b\u662f\u5426\u5173\u6ce8\u89c6\u76d8\u9644\u8fd1\u533a\u57df\u6216\u5176\u4ed6\u56fe\u50cf\u7279\u5f81\u3002<\/p>\n

        \u9700\u8981\u6ce8\u610f，\u70ed\u529b\u56fe\u4ec5\u8868\u793a\u6a21\u578b\u5173\u6ce8\u533a\u57df，\u4e0d\u80fd\u76f4\u63a5\u4f5c\u4e3a\u4e34\u5e8a\u75c5\u7076\u5b9a\u4f4d\u4f9d\u636e\u3002
        \n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
        \n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/p>\n

        \u5341\u4e00\u3001\u5982\u679c\u5f00\u53d1Web\u7cfb\u7edf，\u53ef\u4ee5\u8bbe\u8ba1\u54ea\u4e9b\u529f\u80fd？<\/h3>\n

        \u63a8\u8350\u6280\u672f\u67b6\u6784：<\/p>\n

        PyTorch + ResNet18 + Flask + Vue + ECharts<\/p>\n\n\u529f\u80fd\u6a21\u5757\u529f\u80fd\u8bf4\u660e<\/tr>\n\n\n\n\n\n\n\n\n\n\n\n\n\n
        \u7cfb\u7edf\u9996\u9875<\/td>\n\u5c55\u793a\u6570\u636e\u96c6\u548c\u6a21\u578b\u4fe1\u606f<\/td>\n<\/tr>\n
        \u56fe\u50cf\u4e0a\u4f20<\/td>\n\u4e0a\u4f20\u773c\u5e95\u56fe\u50cf<\/td>\n<\/tr>\n
        \u9752\u5149\u773c\u5206\u7c7b<\/td>\n\u8f93\u51fa\u6a21\u578b\u9884\u6d4b\u7ed3\u679c<\/td>\n<\/tr>\n
        \u7f6e\u4fe1\u5ea6\u5c55\u793a<\/td>\n\u5c55\u793a\u5206\u7c7b\u6982\u7387<\/td>\n<\/tr>\n
        \u6279\u91cf\u8bc6\u522b<\/td>\n\u6279\u91cf\u5904\u7406\u591a\u5f20\u56fe\u7247<\/td>\n<\/tr>\n
        ExpCDR\u5206\u6790<\/td>\n\u5c55\u793a\u676f\u76d8\u6bd4\u6570\u503c\u4e0e\u7edf\u8ba1<\/td>\n<\/tr>\n
        \u6a21\u578b\u5207\u6362<\/td>\n\u5207\u6362ResNet\u3001DenseNet\u7b49\u6a21\u578b<\/td>\n<\/tr>\n
        \u6a21\u578b\u8bc4\u4ef7<\/td>\n\u5c55\u793aAccuracy\u3001Recall\u3001F1\u3001AUC<\/td>\n<\/tr>\n
        \u8bad\u7ec3\u7ed3\u679c<\/td>\n\u5c55\u793aLoss\u548cAccuracy\u66f2\u7ebf<\/td>\n<\/tr>\n
        Grad-CAM<\/td>\n\u5c55\u793a\u6a21\u578b\u5173\u6ce8\u533a\u57df<\/td>\n<\/tr>\n
        \u5386\u53f2\u8bb0\u5f55<\/td>\n\u4fdd\u5b58\u8bc6\u522b\u7ed3\u679c<\/td>\n<\/tr>\n
        \u6570\u636e\u53ef\u89c6\u5316<\/td>\n\u5c55\u793a\u6807\u7b7e\u548c\u6307\u6807\u5206\u5e03<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

        \u7cfb\u7edf\u5de5\u4f5c\u6d41\u7a0b：<\/p>\n

        \u773c\u5e95\u56fe\u50cf\u4e0a\u4f20
        \n \u2193
        \n\u56fe\u50cf\u9884\u5904\u7406
        \n \u2193
        \nResNet \/ DenseNet\u5206\u7c7b\u6a21\u578b
        \n \u2193
        \n\u9752\u5149\u773c\u4e8c\u5206\u7c7b\u9884\u6d4b
        \n \u2193
        \n\u9884\u6d4b\u6982\u7387\u4e0e\u70ed\u529b\u56fe\u5c55\u793a
        \n \u2193
        \n\u7ed3\u679c\u4fdd\u5b58\u4e0e\u53ef\u89c6\u5316<\/p>\n

        \u5982\u679c\u9009\u62e9ExpCDR\u878d\u5408\u65b9\u5411，\u5219\u9700\u8981\u589e\u52a0\u6570\u503c\u8f93\u5165\u548c\u7279\u5f81\u878d\u5408\u6a21\u5757\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

        \u3010\u6df1\u5ea6\u5b66\u4e60\u6570\u636e\u96c6\u63a8\u8350\u3011\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6，\u5305\u542b\u773c\u5e95\u56fe\u50cf\u4e0e650\u6761\u6807\u6ce8\u8bb0\u5f55，\u9644Python\u4ee3\u7801\u53ca\u6bd5\u8bbe\u9898\u76ee
        \n\u4eca\u5929\u5206\u4eab\u4e00\u4e2a\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1\u7684\u516c\u5f00\u6570\u636e\u96c6
        \n\u8be5\u6570\u636e\u96c6\u4e3b\u8981\u7528\u4e8e\u9752\u5149\u773c\u76f8\u5173\u5f71\u50cf\u8bc6\u522b\u7814\u7a76，\u5305\u542bACRIMA\u3001ORIGA\u7b49\u6570\u636e\u76ee\u5f55，\u4ee5\u53ca\u8bb0\u5f55\u56fe\u50cf\u540d\u79f0\u3001\u89c6\u676f\u76d8\u6bd4\u548c\u9752\u5149\u773c\u6807\u7b7e\u7684CSV\u6587\u4ef6\u3002
        \n\u5bf9\u4e8e\u8ba1\u7b97\u673a\u3001\u4eba\u5de5\u667a\u80fd\u3001\u533b\u5b66\u4fe1\u606f\u5de5\u7a0b\u3001\u751f\u7269\u533b\u5b66\u5de5\u7a0b\u7b49\u4e13\u4e1a\u7684\u540c\u5b66，\u53ef\u4ee5\u56f4\u7ed5\u9752\u5149\u773c\u56fe\u50cf\u5206\u7c7b\u3001<\/p>\n","protected":false},"author":2,"featured_media":121442,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[13897,6819,13896,13895,86],"topic":[],"class_list":["post-121447","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-server","tag-mobilenetv3","tag-resnet","tag-13896","tag-13895","tag-86"],"yoast_head":"\n\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/121447.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u3010\u6df1\u5ea6\u5b66\u4e60\u6570\u636e\u96c6\u63a8\u8350\u3011\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6，\u5305\u542b\u773c\u5e95\u56fe\u50cf\u4e0e650\u6761\u6807\u6ce8\u8bb0\u5f55，\u9644Python\u4ee3\u7801\u53ca\u6bd5\u8bbe\u9898\u76ee \u4eca\u5929\u5206\u4eab\u4e00\u4e2a\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1\u7684\u516c\u5f00\u6570\u636e\u96c6 \u8be5\u6570\u636e\u96c6\u4e3b\u8981\u7528\u4e8e\u9752\u5149\u773c\u76f8\u5173\u5f71\u50cf\u8bc6\u522b\u7814\u7a76，\u5305\u542bACRIMA\u3001ORIGA\u7b49\u6570\u636e\u76ee\u5f55，\u4ee5\u53ca\u8bb0\u5f55\u56fe\u50cf\u540d\u79f0\u3001\u89c6\u676f\u76d8\u6bd4\u548c\u9752\u5149\u773c\u6807\u7b7e\u7684CSV\u6587\u4ef6\u3002 \u5bf9\u4e8e\u8ba1\u7b97\u673a\u3001\u4eba\u5de5\u667a\u80fd\u3001\u533b\u5b66\u4fe1\u606f\u5de5\u7a0b\u3001\u751f\u7269\u533b\u5b66\u5de5\u7a0b\u7b49\u4e13\u4e1a\u7684\u540c\u5b66，\u53ef\u4ee5\u56f4\u7ed5\u9752\u5149\u773c\u56fe\u50cf\u5206\u7c7b\u3001\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/121447.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-11-02T12:32:49+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2026\/11\/20261102123248-6ae882f074515.png\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"5 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/121447.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/121447.html\",\"name\":\"\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-11-02T12:32:49+00:00\",\"dateModified\":\"2026-11-02T12:32:49+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/121447.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/121447.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/121447.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/121447.html","og_locale":"zh_CN","og_type":"article","og_title":"\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u3010\u6df1\u5ea6\u5b66\u4e60\u6570\u636e\u96c6\u63a8\u8350\u3011\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6，\u5305\u542b\u773c\u5e95\u56fe\u50cf\u4e0e650\u6761\u6807\u6ce8\u8bb0\u5f55，\u9644Python\u4ee3\u7801\u53ca\u6bd5\u8bbe\u9898\u76ee \u4eca\u5929\u5206\u4eab\u4e00\u4e2a\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1\u7684\u516c\u5f00\u6570\u636e\u96c6 \u8be5\u6570\u636e\u96c6\u4e3b\u8981\u7528\u4e8e\u9752\u5149\u773c\u76f8\u5173\u5f71\u50cf\u8bc6\u522b\u7814\u7a76，\u5305\u542bACRIMA\u3001ORIGA\u7b49\u6570\u636e\u76ee\u5f55，\u4ee5\u53ca\u8bb0\u5f55\u56fe\u50cf\u540d\u79f0\u3001\u89c6\u676f\u76d8\u6bd4\u548c\u9752\u5149\u773c\u6807\u7b7e\u7684CSV\u6587\u4ef6\u3002 \u5bf9\u4e8e\u8ba1\u7b97\u673a\u3001\u4eba\u5de5\u667a\u80fd\u3001\u533b\u5b66\u4fe1\u606f\u5de5\u7a0b\u3001\u751f\u7269\u533b\u5b66\u5de5\u7a0b\u7b49\u4e13\u4e1a\u7684\u540c\u5b66，\u53ef\u4ee5\u56f4\u7ed5\u9752\u5149\u773c\u56fe\u50cf\u5206\u7c7b\u3001","og_url":"https:\/\/www.wsisp.com\/helps\/121447.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-11-02T12:32:49+00:00","og_image":[{"url":"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2026\/11\/20261102123248-6ae882f074515.png"}],"author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"5 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/121447.html","url":"https:\/\/www.wsisp.com\/helps\/121447.html","name":"\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-11-02T12:32:49+00:00","dateModified":"2026-11-02T12:32:49+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/121447.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/121447.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/121447.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"\u6570\u636e\u96c638-\u9752\u5149\u773c\u68c0\u6d4b\u6570\u636e\u96c6\uff0c\u9002\u5408\u533b\u5b66\u5f71\u50cf\u5206\u6790\u3001\u6df1\u5ea6\u5b66\u4e60\u56fe\u50cf\u5206\u7c7b\u3001\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u8ba1\u7b97\u673a\u6bd5\u4e1a\u8bbe\u8ba1"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/121447","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=121447"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/121447\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media\/121442"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=121447"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=121447"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=121447"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=121447"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}