{"id":36582,"date":"2026-01-17T17:26:53","date_gmt":"2026-01-17T09:26:53","guid":{"rendered":"http:\/\/192.168.10.115\/?p=36582"},"modified":"2026-01-19T17:31:18","modified_gmt":"2026-01-19T09:31:18","slug":"2026-01-16-yolo26-%e6%ad%a3%e5%bc%8f%e5%bc%80%e6%ba%90%ef%bc%81%e6%97%a0nms%e6%8e%a8%e7%90%86cpu-%e6%80%a7%e8%83%bd%e6%8f%90%e5%8d%8743%ef%bc%8c%e9%9d%a2%e5%90%91%e8%be%b9%e7%bc%98%e8%a7%86%e8%a7%89a","status":"publish","type":"post","link":"http:\/\/222.128.65.89:18086\/index.php\/2026\/01\/17\/36582\/","title":{"rendered":"2026-01-17 YOLO26 \u6b63\u5f0f\u5f00\u6e90\uff01\u65e0NMS\u63a8\u7406+CPU \u6027\u80fd\u63d0\u534743%\uff0c\u9762\u5411\u8fb9\u7f18\u89c6\u89c9AI\u7684\u65b0\u4e00\u4ee3YOLO\u6a21\u578b"},"content":{"rendered":"\n

\u539f\u6587\u94fe\u63a5\uff1ahttps:\/\/mp.weixin.qq.com\/s\/Sum3Y34wa-ZvSG-iKph6tg<\/a><\/p>\n\n\n\n

\u521a\u521a\uff0cUltralytics \u6b63\u5f0f\u53d1\u5e03 YOLO26\u3002<\/strong><\/p>\n\n\n\n

\u8be5\u6a21\u578b\u6700\u65e9\u5728 YOLO Vision 2025\uff08YV25\uff09\u5927\u4f1a\u4e0a\u9996\u6b21\u4eae\u76f8\uff0c\u91cd\u70b9\u56f4\u7ed5\u771f\u5b9e\u4e16\u754c\u89c6\u89c9\u7cfb\u7edf\u4e2d\u7684\u8bad\u7ec3\u65b9\u5f0f\u3001\u90e8\u7f72\u6548\u7387\u4ee5\u53ca\u89c4\u6a21\u5316\u843d\u5730\u80fd\u529b\u8fdb\u884c\u4e86\u7cfb\u7edf\u5347\u7ea7\u3002<\/p>\n\n\n\n

\u5728\u4f26\u6566\u4e3e\u529e\u7684 YV25 \u5927\u4f1a\u4e0a\uff0cUltralytics\u521b\u59cb\u4eba\u517c\u9996\u5e2d\u6267\u884c\u5b98 Glenn Jocher \u5206\u4eab\u4e86\u8fd9\u4e00\u613f\u666f\uff1a\u201c\u6700\u9876\u5c16\u7684 AI \u6280\u672f\u5f80\u5f80\u88ab\u675f\u4e4b\u9ad8\u9601\uff0c\u5e76\u672a\u5bf9\u5916\u5f00\u653e\u3002\u5927\u578b\u4f01\u4e1a\u638c\u63a7\u7740\u65b0\u6280\u672f\u7684\u53d1\u5c55\uff0c\u5176\u4ed6\u4eba\u53ea\u80fd\u6392\u961f\u7b49\u5f85\u4f7f\u7528\u6743\u9650\u3002Ultralytics \u9009\u62e9\u4e86\u4e00\u6761\u4e0d\u540c\u7684\u9053\u8def\uff0c\u6211\u4eec\u5e0c\u671b AI \u80fd\u771f\u6b63\u638c\u63e1\u5728\u6bcf\u4e2a\u4eba\u624b\u4e2d\u3002\u201d<\/p>\n\n\n\n

\u201c\u5c06AI\u4ece\u4e91\u7aef\u5e26\u5230\u771f\u5b9e\u4e16\u754c\u4e2d\u201d<\/strong><\/p>\n\n\n\n

YOLO26 \u6b63\u662f\u56f4\u7ed5\u8fd9\u4e9b\u5b9e\u9645\u9700\u6c42\u8fdb\u884c\u8bbe\u8ba1\uff0c\u5728 CPU\u3001\u8fb9\u7f18\u52a0\u901f\u5668\u4ee5\u53ca\u4f4e\u529f\u8017\u786c\u4ef6\u4e0a\u5b9e\u73b0\u4e86\u517c\u987e\u6027\u80fd\u4e0e\u6548\u7387\u7684\u8fd0\u884c\u8868\u73b0\u3002<\/p>\n\n\n\n

\u5728\u5b9e\u73b0\u67b6\u6784\u4e0e\u6027\u80fd\u5347\u7ea7\u7684\u540c\u65f6\uff0cYOLO26 \u4ecd\u5ef6\u7eed\u4e86 YOLO \u7cfb\u5217\u4e00\u8d2f\u7684\u5de5\u7a0b\u53cb\u597d\u7279\u6027\uff0c\u80fd\u591f\u8f83\u4e3a\u987a\u7545\u5730\u878d\u5165\u73b0\u6709\u5de5\u4f5c\u6d41\u7a0b\uff0c\u652f\u6301\u591a\u7c7b\u89c6\u89c9\u4efb\u52a1\u3002\u65e0\u8bba\u662f\u7814\u7a76\u56e2\u961f\u8fd8\u662f\u751f\u4ea7\u56e2\u961f\u90fd\u80fd\u8f7b\u677e\u4e0a\u624b<\/strong>\u3002<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u56fe 1\uff1aYOLO26 \u7684\u76ee\u6807\u68c0\u6d4b\u5e94\u7528\u793a\u4f8b<\/p>\n\n\n\n

\u672c\u6587\u5c06\u56f4\u7ed5 YOLO26 \u7684\u6838\u5fc3\u8bbe\u8ba1\u601d\u8def\uff0c\u5173\u952e\u6280\u672f\u6539\u8fdb\u4ee5\u53ca\u5176\u5728\u8fb9\u7f18\u89c6\u89c9 AI \u573a\u666f\u4e2d\u7684\u610f\u4e49\u5c55\u5f00\u4ecb\u7ecd\u3002<\/p>\n\n\n\n

Github\u5df2\u5f00\u6e90\uff1a<\/strong> https:\/\/github.com\/ultralytics\/ultralytics<\/p>\n\n\n\n

01 YOLO26 \u7684\u6838\u5fc3\u4f18\u52bf<\/strong><\/p>\n\n\n\n

\u4e0e\u6b64\u524d\u7684 Ultralytics YOLO \u7cfb\u5217\u4e00\u6837\uff0cYOLO26 \u4f9d\u7136\u91c7\u7528\u7edf\u4e00\u7684\u6a21\u578b\u5bb6\u65cf\u8bbe\u8ba1\uff0c\u80fd\u591f\u652f\u6301\u591a\u79cd\u8ba1\u7b97\u673a\u89c6\u89c9\u4efb\u52a1\u3002\u6a21\u578b\u63d0\u4f9b\u4e94\u79cd\u5c3a\u5bf8\u9009\u62e9\uff1aNano\uff08n\uff09\u3001Small\uff08s\uff09\u3001Medium\uff08m\uff09\u3001Large\uff08l\uff09\u548c Extra Large\uff08x\uff09<\/strong>\uff0c\u56e2\u961f\u53ef\u4ee5\u6839\u636e\u90e8\u7f72\u73af\u5883\uff0c\u5728\u901f\u5ea6\u3001\u7cbe\u5ea6\u548c\u6a21\u578b\u4f53\u91cf\u4e4b\u95f4\u8fdb\u884c\u7075\u6d3b\u53d6\u820d\u3002<\/p>\n\n\n\n

\u5728\u6027\u80fd\u5c42\u9762,YOLO26 \u8fdb\u4e00\u6b65\u63d0\u5347\u4e86\u6027\u80fd\u4e0a\u9650\u3002\u4e0e YOLO11 \u76f8\u6bd4\uff0cYOLO26 \u7684 Nano \u7248\u672c\u5728 CPU \u63a8\u7406\u573a\u666f\u4e0b\u6700\u9ad8\u53ef\u5b9e\u73b0 43% \u7684\u6027\u80fd\u63d0\u5347\uff0c\u6210\u4e3a\u76ee\u524d\u8fb9\u7f18\u7aef\u548c\u57fa\u4e8e CPU \u90e8\u7f72\u573a\u666f\u4e2d\u901f\u5ea6\u4e0e\u7cbe\u5ea6\u517c\u987e\u7684\u9886\u5148\u76ee\u6807\u68c0\u6d4b\u6a21\u578b\u4e4b\u4e00\u3002<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\u56fe 2. YOLO26 \u6a21\u578b\u6548\u679c\u5bf9\u6bd4<\/p>\n\n\n\n

YOLO26 \u652f\u6301\u7684\u8ba1\u7b97\u673a\u89c6\u89c9\u4efb\u52a1\u5305\u62ec\uff1a<\/strong><\/h3>\n\n\n\n
    \n
  • \u56fe\u50cf\u5206\u7c7b<\/strong>\uff1a\u5bf9\u6574\u5f20\u56fe\u50cf\u8fdb\u884c\u5206\u6790\uff0c\u5e76\u5f52\u7c7b\u5230\u6307\u5b9a\u7c7b\u522b\uff0c\u5e2e\u52a9\u7cfb\u7edf\u7406\u89e3\u573a\u666f\u6574\u4f53\u8bed\u4e49\u3002<\/li>\n\n\n\n
  • \u76ee\u6807\u68c0\u6d4b<\/strong>\uff1a\u5728\u56fe\u50cf\u6216\u89c6\u9891\u4e2d\u5b9a\u4f4d\u5e76\u8bc6\u522b\u591a\u4e2a\u76ee\u6807\u3002<\/li>\n\n\n\n
  • \u5b9e\u4f8b\u5206\u5272<\/strong>\uff1a\u80fd\u4ee5\u50cf\u7d20\u7ea7\u7cbe\u5ea6\u52fe\u52d2\u5355\u4e2a\u7269\u4f53\u8f6e\u5ed3\u3002<\/li>\n\n\n\n
  • \u59ff\u6001\u4f30\u8ba1<\/strong>\uff1a\u8bc6\u522b\u5173\u952e\u70b9\u5e76\u4f30\u8ba1\u4eba\u4f53\u6216\u5176\u4ed6\u5bf9\u8c61\u7684\u59ff\u6001\u3002<\/li>\n\n\n\n
  • \u65cb\u8f6c\u6846\uff08OBB\uff09\u68c0\u6d4b<\/strong>\uff1a\u652f\u6301\u4e0d\u540c\u89d2\u5ea6\u76ee\u6807\u68c0\u6d4b\uff0c\u9002\u7528\u4e8e\u822a\u62cd\u548c\u536b\u661f\u5f71\u50cf\u3002<\/li>\n\n\n\n
  • \u76ee\u6807\u8ddf\u8e2a<\/strong>\uff1a\u7ed3\u5408 Ultralytics Python \u5305\uff0c\u53ef\u5728\u89c6\u9891\u6d41\u4e2d\u6301\u7eed\u8ddf\u8e2a\u76ee\u6807\u3002<\/li>\n<\/ul>\n\n\n\n

    \u6240\u6709\u4efb\u52a1\u5728\u540c\u4e00\u6846\u67b6\u4e0b\u652f\u6301\u8bad\u7ec3\u3001\u9a8c\u8bc1\u3001\u63a8\u7406\u4e0e\u6a21\u578b\u5bfc\u51fa\u3002<\/p>\n\n\n\n

    02 YOLO26 \u7684\u5173\u952e\u521b\u65b0<\/strong><\/p>\n\n\n\n

    YOLO26 \u5f15\u5165\u4e86\u591a\u9879\u6838\u5fc3\u521b\u65b0\uff0c\u5168\u9762\u63d0\u5347\u4e86\u63a8\u7406\u901f\u5ea6\u3001\u8bad\u7ec3\u7a33\u5b9a\u6027\u548c\u90e8\u7f72\u4fbf\u6377\u6027\uff0c\u4e3b\u8981\u5305\u62ec\uff1a<\/p>\n\n\n\n

      \n
    • \u79fb\u9664\u5206\u5e03\u7126\u70b9\u635f\u5931\uff08DFL\uff09<\/strong>\uff1a\u5220\u9664\u4e86 DFL \u6a21\u5757\uff0c\u7b80\u5316\u4e86\u8fb9\u754c\u6846\u9884\u6d4b\u8fc7\u7a0b\uff0c\u63d0\u5347\u4e86\u786c\u4ef6\u517c\u5bb9\u6027\uff0c\u8ba9\u6a21\u578b\u66f4\u6613\u4e8e\u5bfc\u51fa\u5e76\u5728\u8fb9\u7f18\u8bbe\u5907\u548c\u4f4e\u529f\u8017\u786c\u4ef6\u4e0a\u8fd0\u884c\u3002<\/li>\n\n\n\n
    • \u7aef\u5230\u7aef\u65e0 NMS \u63a8\u7406<\/strong>\uff1aYOLO26 \u539f\u751f\u652f\u6301\u7aef\u5230\u7aef\u63a8\u7406\uff0c\u53ef\u76f4\u63a5\u8f93\u51fa\u6700\u7ec8\u9884\u6d4b\u7ed3\u679c\uff0c\u65e0\u9700\u4f9d\u8d56\u975e\u6781\u5927\u503c\u6291\u5236\uff08NMS\uff09\uff0c\u51cf\u5c11\u4e86\u63a8\u7406\u5ef6\u8fdf\u548c\u90e8\u7f72\u590d\u6742\u5ea6\u3002<\/li>\n\n\n\n
    • \u6e10\u8fdb\u5f0f\u635f\u5931\u5e73\u8861\uff08ProgLoss\uff09+ \u5c0f\u76ee\u6807\u611f\u77e5\u6807\u7b7e\u5206\u914d\uff08STAL\uff09<\/strong>\uff1a\u8fd9\u4e9b\u4f18\u5316\u7684\u635f\u5931\u7b56\u7565\u7a33\u5b9a\u4e86\u8bad\u7ec3\u8fc7\u7a0b\uff0c\u63d0\u5347\u4e86\u68c0\u6d4b\u7cbe\u5ea6\uff0c\u5c24\u5176\u9488\u5bf9\u96be\u4ee5\u68c0\u6d4b\u7684\u5c0f\u76ee\u6807\u3002<\/li>\n\n\n\n
    • MuSGD \u4f18\u5316\u5668<\/strong>\uff1a\u91c7\u7528\u5168\u65b0\u6df7\u5408\u4f18\u5316\u5668\uff0c\u878d\u5408\u4e86 SGD \u4e0e Muon \u7684\u4f18\u5316\u65b9\u6cd5\uff0c\u6a21\u578b\u8bad\u7ec3\u66f4\u52a0\u7a33\u5b9a\u3002<\/li>\n\n\n\n
    • CPU \u63a8\u7406\u901f\u5ea6\u63d0\u5347\u9ad8\u8fbe 43%<\/strong>\u00a0\uff1a\u4e13\u4e3a\u8fb9\u7f18\u8ba1\u7b97\u4f18\u5316\uff0cYOLO26 \u7684 CPU \u63a8\u7406\u901f\u5ea6\u63d0\u5347\u9ad8\u8fbe 43%\uff0c\u53ef\u5728\u8fb9\u7f18\u8bbe\u5907\u4e0a\u5b9e\u73b0\u5b9e\u65f6\u6027\u80fd\u3002<\/li>\n<\/ul>\n\n\n\n

      \u63a5\u4e0b\u6765\uff0c\u6211\u4eec\u5c06\u8be6\u7ec6\u62c6\u89e3\u8fd9\u4e9b\u4e0b\u4e00\u4ee3\u7279\u6027\u3002<\/p>\n\n\n\n

      2.1 \u79fb\u9664\u5206\u5e03\u7126\u70b9\u635f\u5931\uff08DFL\uff09\uff0c\u7b80\u5316\u9884\u6d4b\u6d41\u7a0b<\/strong><\/p>\n\n\n\n

      \u65e9\u671f YOLO \u6a21\u578b\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u4f7f\u7528\u5206\u5e03\u7126\u70b9\u635f\u5931\uff08DFL\uff09\u6765\u63d0\u5347\u8fb9\u754c\u6846\u7cbe\u5ea6\u3002\u5c3d\u7ba1\u8fd9\u4e00\u65b9\u6cd5\u6709\u6548\uff0c\u4f46 DFL \u589e\u52a0\u4e86\u6a21\u578b\u590d\u6742\u5ea6\uff0c\u4e14\u5bf9\u56de\u5f52\u8303\u56f4\u8bbe\u7f6e\u4e86\u56fa\u5b9a\u9650\u5236\uff0c\u7ed9\u6a21\u578b\u5bfc\u51fa\u548c\u90e8\u7f72\u5e26\u6765\u6311\u6218\uff0c\u5c24\u5176\u5728\u8fb9\u7f18\u8bbe\u5907\u548c\u4f4e\u529f\u8017\u786c\u4ef6\u4e0a\u8868\u73b0\u66f4\u4e3a\u660e\u663e\u3002<\/p>\n\n\n\n

      YOLO26 \u5b8c\u5168\u79fb\u9664\u4e86 DFL \u6a21\u5757\u3002\u8fd9\u4e00\u6539\u52a8\u6d88\u9664\u4e86\u65e9\u671f\u6a21\u578b\u4e2d\u56fa\u5b9a\u7684\u8fb9\u754c\u6846\u56de\u5f52\u9650\u5236\uff0c\u63d0\u5347\u4e86\u68c0\u6d4b\u8d85\u5927\u7269\u4f53\u65f6\u7684\u53ef\u9760\u6027\u548c\u51c6\u786e\u6027\u3002<\/p>\n\n\n\n

      \u901a\u8fc7\u7b80\u5316\u8fb9\u754c\u6846\u9884\u6d4b\u6d41\u7a0b\uff0cYOLO26 \u66f4\u6613\u4e8e\u5bfc\u51fa\uff0c\u4e14\u80fd\u5728\u5404\u7c7b\u8fb9\u7f18\u8bbe\u5907\u548c\u4f4e\u529f\u8017\u786c\u4ef6\u4e0a\u7a33\u5b9a\u8fd0\u884c\u3002<\/p>\n\n\n\n

      2.2 \u7aef\u5230\u7aef\u65e0 NMS \u63a8\u7406<\/strong><\/p>\n\n\n\n

      \u4f20\u7edf\u76ee\u6807\u68c0\u6d4b\u6d41\u7a0b\u4f9d\u8d56\u975e\u6781\u5927\u503c\u6291\u5236\uff08NMS\uff09\u4f5c\u4e3a\u540e\u5904\u7406\u6b65\u9aa4\uff0c\u7528\u4e8e\u8fc7\u6ee4\u91cd\u53e0\u9884\u6d4b\u7ed3\u679c\u3002\u5c3d\u7ba1 NMS \u6548\u679c\u663e\u8457\uff0c\u4f46\u5b83\u4f1a\u589e\u52a0\u5ef6\u8fdf\u548c\u590d\u6742\u5ea6\uff0c\u4e14\u5728\u591a\u8fd0\u884c\u65f6\u73af\u5883\u548c\u4e0d\u540c\u786c\u4ef6\u76ee\u6807\u4e0a\u90e8\u7f72\u65f6\u5bb9\u6613\u51fa\u73b0\u4e0d\u7a33\u5b9a\u95ee\u9898\u3002<\/p>\n\n\n\n

      YOLO26 \u5f15\u5165\u539f\u751f\u7aef\u5230\u7aef\u63a8\u7406\u6a21\u5f0f\uff0c\u6a21\u578b\u53ef\u76f4\u63a5\u8f93\u51fa\u6700\u7ec8\u9884\u6d4b\u7ed3\u679c\uff0c\u65e0\u9700\u5c06 NMS \u4f5c\u4e3a\u72ec\u7acb\u7684\u540e\u5904\u7406\u6b65\u9aa4\u3002\u91cd\u590d\u9884\u6d4b\u7684\u8fc7\u6ee4\u7684\u64cd\u4f5c\u5728\u7f51\u7edc\u5185\u90e8\u5b8c\u6210\u3002<\/p>\n\n\n\n

      \u79fb\u9664 NMS \u4e0d\u4ec5\u964d\u4f4e\u4e86\u5ef6\u8fdf\uff0c\u7b80\u5316\u4e86\u90e8\u7f72\u6d41\u7a0b\uff0c\u8fd8\u51cf\u5c11\u4e86\u96c6\u6210\u9519\u8bef\u7684\u98ce\u9669\uff0c\u4f7f YOLO26 \u7279\u522b\u9002\u5408\u5b9e\u65f6\u90e8\u7f72\u548c\u8fb9\u7f18\u90e8\u7f72\u573a\u666f\u3002<\/p>\n\n\n\n

      2.3 \u6e10\u8fdb\u5f0f\u635f\u5931\u5e73\u8861\uff08ProgLoss\uff09+ \u5c0f\u76ee\u6807\u611f\u77e5\u6807\u7b7e\u5206\u914d\uff08STAL\uff09\uff0c\u63d0\u5347\u8bc6\u522b\u80fd\u529b<\/strong><\/p>\n\n\n\n

      \u8bad\u7ec3\u76f8\u5173\u7684\u5173\u952e\u7279\u6027\u662f\u5f15\u5165\u4e86\u6e10\u8fdb\u5f0f\u635f\u5931\u5e73\u8861\uff08ProgLoss\uff09\u548c\u5c0f\u76ee\u6807\u611f\u77e5\u6807\u7b7e\u5206\u914d\uff08STAL\uff09\u3002\u8fd9\u4e9b\u4f18\u5316\u7684\u635f\u5931\u51fd\u6570\u6709\u52a9\u4e8e\u7a33\u5b9a\u8bad\u7ec3\u8fc7\u7a0b\uff0c\u63d0\u5347\u68c0\u6d4b\u7cbe\u5ea6\u3002<\/p>\n\n\n\n

      ProgLoss \u8ba9\u6a21\u578b\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u5b66\u4e60\u66f4\u7a33\u5b9a\uff0c\u51cf\u5c11\u4e86\u8bad\u7ec3\u6ce2\u52a8\uff0c\u4f7f\u6a21\u578b\u6536\u655b\u66f4\u5e73\u7a33\u3002\u540c\u65f6\uff0cSTAL \u5219\u9488\u5bf9\u5c0f\u76ee\u6807\u8fdb\u884c\u4e86\u4f18\u5316\uff0c\u6539\u5584\u6a21\u578b\u5728\u89c6\u89c9\u4fe1\u606f\u6709\u9650\u60c5\u51b5\u4e0b\u7684\u5b66\u4e60\u6548\u679c\u3002<\/p>\n\n\n\n

      ProgLoss \u548c STAL \u7684\u7ed3\u5408\u5b9e\u73b0\u4e86\u66f4\u53ef\u9760\u7684\u68c0\u6d4b\u6548\u679c\uff0c\u5c24\u5176\u5728\u5c0f\u76ee\u6807\u8bc6\u522b\u65b9\u9762\u6709\u663e\u8457\u63d0\u5347\u3002\u8fd9\u4e00\u70b9\u5bf9\u4e8e\u7269\u8054\u7f51\uff08IoT\uff09\u3001\u673a\u5668\u4eba\u6280\u672f\u548c\u822a\u7a7a\u5f71\u50cf\u7b49\u8fb9\u7f18\u5e94\u7528\u81f3\u5173\u91cd\u8981\uff0c\u56e0\u4e3a\u5728\u8fd9\u4e9b\u573a\u666f\u4e2d\uff0c\u7269\u4f53\u5f80\u5f80\u4f53\u79ef\u5c0f\u3001\u8ddd\u79bb\u8fdc\u6216\u90e8\u5206\u53ef\u89c1\u3002<\/p>\n\n\n\n

      2.4 MuSGD \u4f18\u5316\u5668\uff0c\u5b9e\u73b0\u66f4\u7a33\u5b9a\u7684\u8bad\u7ec3<\/strong><\/p>\n\n\n\n

      YOLO26 \u91c7\u7528\u4e86\u4e00\u79cd\u65b0\u7684\u4f18\u5316\u5668 MuSGD\uff0c\u7528\u4e8e\u63d0\u5347\u8bad\u7ec3\u7684\u7a33\u5b9a\u6027\u548c\u6548\u7387\u3002MuSGD \u5c06\u4f20\u7edf\u968f\u673a\u68af\u5ea6\u4e0b\u964d\uff08SGD\uff09\u7684\u4f18\u52bf\uff0c\u4e0e\u6e90\u81ea\u5927\u8bed\u8a00\u6a21\u578b\u8bad\u7ec3\u7684 Muon \u7c7b\u4f18\u5316\u601d\u60f3\u76f8\u7ed3\u5408\u3002<\/p>\n\n\n\n

      SGD \u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u9886\u57df\u957f\u671f\u88ab\u8bc1\u660e\u5177\u6709\u826f\u597d\u7684\u6cdb\u5316\u80fd\u529b\uff0c\u800c\u8fd1\u5e74\u7684\u5927\u6a21\u578b\u8bad\u7ec3\u7ecf\u9a8c\u4e5f\u8868\u660e\uff0c\u9002\u5f53\u5f15\u5165\u65b0\u7684\u4f18\u5316\u7b56\u7565\u53ef\u4ee5\u8fdb\u4e00\u6b65\u63d0\u5347\u7a33\u5b9a\u6027\u548c\u6548\u7387\u3002MuSGD \u5c06\u8fd9\u4e9b\u7406\u5ff5\u5f15\u5165\u4e86\u8ba1\u7b97\u673a\u89c6\u89c9\u9886\u57df\u3002<\/p>\n\n\n\n

      \u53d7 Moonshot AI \u7684 Kimi K2 \u8bad\u7ec3\u7ecf\u9a8c\u542f\u53d1\uff0cMuSGD \u878d\u5165\u4e86\u6709\u52a9\u4e8e\u6a21\u578b\u66f4\u5e73\u7a33\u6536\u655b\u7684\u4f18\u5316\u7b56\u7565\u3002\u8fd9\u4f7f\u5f97 YOLO26 \u80fd\u591f\u66f4\u5feb\u5730\u8fbe\u5230\u51fa\u8272\u6027\u80fd\uff0c\u540c\u65f6\u51cf\u5c11\u8bad\u7ec3\u4e0d\u7a33\u5b9a\u6027\uff0c\u5c24\u5176\u5728\u66f4\u5927\u89c4\u6a21\u6216\u66f4\u590d\u6742\u7684\u8bad\u7ec3\u573a\u666f\u4e2d\u6548\u679c\u663e\u8457\u3002<\/p>\n\n\n\n

      MuSGD \u8ba9 YOLO26 \u5728\u4e0d\u540c\u6a21\u578b\u5c3a\u5bf8\u4e0b\u7684\u8bad\u7ec3\u90fd\u66f4\u5177\u53ef\u9884\u6d4b\u6027\uff0c\u65e2\u63d0\u5347\u4e86\u6027\u80fd\uff0c\u53c8\u589e\u5f3a\u4e86\u8bad\u7ec3\u7a33\u5b9a\u6027\u3002<\/p>\n\n\n\n

      2.5 CPU \u63a8\u7406\u901f\u5ea6\u63d0\u5347\u9ad8\u8fbe 43%<\/strong><\/p>\n\n\n\n

      \u968f\u7740 Vision AI \u4e0d\u65ad\u5411\u6570\u636e\u4ea7\u751f\u7aef\u9760\u8fd1\uff0c\u8fb9\u7f18\u6027\u80fd\u7684\u91cd\u8981\u6027\u6301\u7eed\u63d0\u5347\u3002YOLO26 \u9488\u5bf9\u8fb9\u7f18\u8ba1\u7b97\u573a\u666f\u8fdb\u884c\u4e86\u6df1\u5ea6\u4f18\u5316\uff0c\u5728\u65e0 GPU \u7684\u6761\u4ef6\u4e0b\uff0cCPU \u63a8\u7406\u901f\u5ea6\u6700\u9ad8\u63d0\u5347\u53ef\u8fbe 43%\u3002<\/p>\n\n\n\n

      \u8fd9\u4e00\u80fd\u529b\u4f7f\u5b9e\u65f6\u89c6\u89c9\u7cfb\u7edf\u80fd\u591f\u76f4\u63a5\u8fd0\u884c\u5728\u6444\u50cf\u5934\u3001\u673a\u5668\u4eba\u548c\u5d4c\u5165\u5f0f\u786c\u4ef6\u4e0a\uff0c\u6ee1\u8db3\u4f4e\u5ef6\u8fdf\u3001\u9ad8\u53ef\u9760\u6027\u548c\u6210\u672c\u53d7\u9650\u7684\u5b9e\u9645\u9700\u6c42\u3002<\/p>\n\n\n\n

      2.6 \u5bf9\u89c6\u89c9\u4efb\u52a1\u8fdb\u4e00\u6b65\u4f18\u5316<\/strong><\/p>\n\n\n\n

      \u9664\u4e86\u63d0\u5347\u76ee\u6807\u68c0\u6d4b\u7cbe\u5ea6\u7684\u67b6\u6784\u6539\u8fdb\u5916\uff0cYOLO26 \u8fd8\u5305\u542b\u9488\u5bf9\u7279\u5b9a\u4efb\u52a1\u7684\u4f18\u5316\uff0c\u65e8\u5728\u5168\u9762\u63d0\u5347\u5404\u7c7b\u8ba1\u7b97\u673a\u89c6\u89c9\u4efb\u52a1\u7684\u6027\u80fd\u3002<\/p>\n\n\n\n

        \n
      • \u5b9e\u4f8b\u5206\u5272<\/strong>\uff1aYOLO26 \u4f7f\u7528\u8bed\u4e49\u5206\u5272\u635f\u5931\u6539\u8fdb\u6a21\u578b\u7684\u8bad\u7ec3\u5b66\u4e60\u8fc7\u7a0b\uff0c\u4ece\u800c\u751f\u6210\u66f4\u51c6\u786e\u3001\u66f4\u4e00\u81f4\u7684\u5b9e\u4f8b\u63a9\u7801\u3002\u5347\u7ea7\u540e\u7684\u539f\u578b\u6a21\u5757\u652f\u6301\u5229\u7528\u591a\u5c3a\u5ea6\u4fe1\u606f\uff0c\u4f7f\u6a21\u578b\u5728\u590d\u6742\u573a\u666f\u4e2d\u4e5f\u80fd\u6709\u6548\u5904\u7406\u4e0d\u540c\u5c3a\u5bf8\u7684\u7269\u4f53\u3002<\/li>\n\n\n\n
      • \u59ff\u6001\u4f30\u8ba1<\/strong>\uff1a\u96c6\u6210 Residual Log-Likelihood Estimation\uff08RLE\uff09\uff0c\u66f4\u597d\u5730\u5efa\u6a21\u5173\u952e\u70b9\u9884\u6d4b\u7684\u4e0d\u786e\u5b9a\u6027\uff0c\u540c\u65f6\u4f18\u5316\u89e3\u7801\u6d41\u7a0b\uff0c\u5b9e\u73b0\u66f4\u9ad8\u7cbe\u5ea6\u4e0e\u5b9e\u65f6\u6027\u3002<\/li>\n\n\n\n
      • \u65cb\u8f6c\u8fb9\u754c\u6846\u68c0\u6d4b<\/strong>\uff1a\u65b0\u589e\u89d2\u5ea6\u635f\u5931\u51fd\u6570\uff0c\u5e2e\u52a9\u6a21\u578b\u66f4\u51c6\u786e\u5730\u5b66\u4e60\u7269\u4f53\u65cb\u8f6c\u89d2\u5ea6\uff0c\u5c24\u5176\u9002\u7528\u4e8e\u65b9\u5411\u6a21\u7cca\u7684\u65b9\u5f62\u7269\u4f53\u3002\u4f18\u5316\u540e\u7684 OBB \u89e3\u7801\u8fc7\u7a0b\u51cf\u5c11\u4e86\u65cb\u8f6c\u8fb9\u754c\u9644\u8fd1\u89d2\u5ea6\u9884\u6d4b\u7684\u7a81\u53d8\uff0c\u5b9e\u73b0\u4e86\u66f4\u7a33\u5b9a\u3001\u4e00\u81f4\u7684\u65b9\u5411\u4f30\u8ba1\u3002<\/li>\n<\/ul>\n\n\n\n
        \"\"<\/figure>\n\n\n\n

        \u56fe 3. YOLO26 \u7684\u5b9e\u4f8b\u5206\u5272\u5e94\u7528\u793a\u4f8b<\/p>\n\n\n\n

        03 YOLOE-26\uff1a\u57fa\u4e8e YOLO26 \u7684\u5f00\u653e\u8bcd\u6c47\u5206\u5272\u6a21\u578b<\/strong><\/p>\n\n\n\n

        Ultralytics \u540c\u65f6\u63a8\u51fa\u4e86 YOLOE-26\u7cfb\u5217\uff1a\u4e00\u4e2a\u57fa\u4e8e YOLO26 \u67b6\u6784\u548c\u8bad\u7ec3\u7b56\u7565\u7684\u5f00\u653e\u8bcd\u6c47\u5206\u5272\u6a21\u578b\u3002<\/p>\n\n\n\n

        YOLOE-26 \u5e76\u975e\u5168\u65b0\u4efb\u52a1\u6216\u529f\u80fd\uff0c\u800c\u662f\u4e00\u4e2a\u4e13\u7528\u6a21\u578b\u5bb6\u65cf\uff0c\u5b83\u590d\u7528\u4e86\u73b0\u6709\u7684\u5206\u5272\u4efb\u52a1\uff0c\u540c\u65f6\u652f\u6301\u6587\u672c\u63d0\u793a\u3001\u89c6\u89c9\u63d0\u793a\u548c\u65e0\u63d0\u793a\u63a8\u7406\u3002\u8be5\u6a21\u578b\u5bb6\u65cf\u6db5\u76d6\u6240\u6709\u6807\u51c6 YOLO \u5c3a\u5bf8\uff0c\u76f8\u6bd4\u65e9\u671f\u5f00\u653e\u8bcd\u6c47\u5206\u5272\u6a21\u578b\uff0cYOLOE-26 \u7684\u7cbe\u5ea6\u66f4\u9ad8\uff0c\u5b9e\u9645\u5e94\u7528\u6027\u80fd\u66f4\u53ef\u9760\u3002<\/p>\n\n\n\n

        04 \u652f\u6301\u591a\u5e73\u53f0\u3001\u591a\u786c\u4ef6\u7684\u5b9e\u9645\u90e8\u7f72<\/strong><\/p>\n\n\n\n

        \u5728\u5b9e\u9645\u5e94\u7528\u4e2d\uff0cYOLO26 \u53ef\u8f7b\u677e\u90e8\u7f72\u5728\u5404\u7c7b\u786c\u4ef6\u4e0a<\/strong>\u3002\u901a\u8fc7 Ultralytics \u63d0\u4f9b\u7684 Python \u5de5\u5177\u53ca\u76f8\u5173\u96c6\u6210\u80fd\u529b\uff0c\u6a21\u578b\u53ef\u4ee5\u5bfc\u51fa\u4e3a\u9762\u5411\u4e0d\u540c\u5e73\u53f0\u548c\u786c\u4ef6\u52a0\u901f\u5668\u7684\u63a8\u7406\u683c\u5f0f\u3002<\/p>\n\n\n\n

        \u4f8b\u5982\uff0c\u5bfc\u51fa\u4e3a TensorRT \u683c\u5f0f\u53ef\u5728 NVIDIA GPU \u4e0a\u5b9e\u73b0\u9ad8\u6027\u80fd\u63a8\u7406\uff0cCoreML \u683c\u5f0f\u652f\u6301\u82f9\u679c\u8bbe\u5907\u539f\u751f\u90e8\u7f72\uff0cOpenVINO \u5219\u80fd\u4f18\u5316\u82f1\u7279\u5c14\u786c\u4ef6\u7684\u8fd0\u884c\u6027\u80fd\u3002YOLO26 \u8fd8\u53ef\u5bfc\u51fa\u81f3\u591a\u6b3e\u4e13\u7528\u8fb9\u7f18\u52a0\u901f\u5668\uff0c\u5728\u4e13\u4e1a\u8fb9\u7f18 AI \u786c\u4ef6\u4e0a\u5b9e\u73b0\u9ad8\u541e\u5410\u91cf\u3001\u9ad8\u80fd\u6548\u6bd4\u7684\u63a8\u7406<\/strong>\u3002<\/p>\n\n\n\n

        05 \u5404\u884c\u4e1a\u8ba1\u7b97\u673a\u89c6\u89c9\u573a\u666f\u5e94\u7528<\/strong><\/p>\n\n\n\n

        YOLO26 \u4ee5\u5b9e\u9645\u90e8\u7f72\u4e3a\u6838\u5fc3\u8bbe\u8ba1\u76ee\u6807\uff0c\u53ef\u5e94\u7528\u4e8e\u591a\u7c7b\u884c\u4e1a\u573a\u666f\uff0c\u5305\u62ec\uff1a<\/p>\n\n\n\n

          \n
        • \u673a\u5668\u4eba\uff1a\u5bfc\u822a\u3001\u969c\u788d\u7269\u68c0\u6d4b\u4e0e\u4ea4\u4e92<\/li>\n\n\n\n
        • \u5236\u9020\u4e1a\uff1a\u7f3a\u9677\u68c0\u6d4b\u3001\u6d41\u7a0b\u76d1\u63a7<\/li>\n\n\n\n
        • \u65e0\u4eba\u673a\u4e0e\u822a\u7a7a\u5f71\u50cf\uff1a\u822a\u62cd\u5206\u6790\u3001\u6d4b\u7ed8\u4e0e\u52d8\u6d4b<\/li>\n\n\n\n
        • \u5d4c\u5165\u5f0f\u4e0e IoT\uff1a\u667a\u80fd\u6444\u50cf\u5934\u3001\u672c\u5730\u89c6\u89c9\u5904\u7406<\/li>\n\n\n\n
        • \u667a\u6167\u57ce\u5e02\uff1a\u4ea4\u901a\u76d1\u63a7\u3001\u516c\u5171\u5b89\u5168\u4e0e\u57fa\u7840\u8bbe\u65bd\u7ba1\u7406<\/li>\n<\/ul>\n\n\n\n
          \"\"<\/figure>\n\n\n\n

          \u56fe 4\uff1aYOLO26 \u7684\u5178\u578b\u5e94\u7528\u573a\u666f<\/p>\n\n\n\n

          06 \u5feb\u901f\u4e0a\u624b \u00a0YOLO26<\/strong><\/p>\n\n\n\n

          YOLO26 \u63d0\u4f9b\u4e24\u79cd\u4e3b\u8981\u4f7f\u7528\u65b9\u5f0f\uff0c\u65e2\u652f\u6301\u5e73\u53f0\u5316\u4f7f\u7528\uff0c\u4e5f\u53ef\u901a\u8fc7\u5f00\u6e90\u6d41\u7a0b\u8fdb\u884c\u96c6\u6210\u3002<\/p>\n\n\n\n

          \u5f00\u53d1\u8005\u53ef\u57fa\u4e8e Python \u5de5\u4f5c\u6d41\u5b8c\u6210\u6a21\u578b\u52a0\u8f7d\u3001\u63a8\u7406\u4e0e\u5bfc\u51fa\uff0c\u5e76\u90e8\u7f72\u81f3 ONNX\u3001TensorRT\u3001CoreML\u3001OpenVINO \u7b49\u5e38\u89c1\u8fd0\u884c\u73af\u5883\u3002<\/p>\n\n\n\n

          ip install ultralytics<\/code><\/pre>\n\n\n\n
          from ultralytics import YOLO  

          # \u52a0\u8f7dCOCO\u9884\u8bad\u7ec3\u7684YOLO26n\u6a21\u578b  <\/em>
          model = YOLO(\"yolo26n.pt\")  

          # \u4f7f\u7528YOLO26n\u6a21\u578b\u5bf9'bus.jpg'\u56fe\u50cf\u8fdb\u884c\u63a8\u7406  <\/em>
          results = model(\"path\/to\/bus.jpg\")<\/code><\/pre>\n\n\n\n
          \u5b8c\u6574\u7684\u4f7f\u7528\u6307\u5357\u548c\u6559\u7a0b\u53ef\u4ee5\u5728Ultralytics \u5b98\u65b9\u6587\u6863\u4e2d\u67e5\u770b\uff08https:\/\/docs.ultralytics.com\/\uff09<\/p>\n\n\n\n
          \u603b\u7ed3<\/strong><\/p>\n\n\n\n
          \u5c0f\u7f16\u8ba4\u4e3a\uff0cYOLO26 \u5e76\u4e0d\u662f\u5355\u7eaf\u8ffd\u6c42\u53c2\u6570\u6216\u699c\u5355\u6210\u7ee9\u7684\u5347\u7ea7\uff0c\u800c\u662f\u628a\u91cd\u5fc3\u653e\u5728\u4e86\u6a21\u578b\u5982\u4f55\u771f\u6b63\u8dd1\u8d77\u6765\u8fd9\u4ef6\u4e8b\u4e0a\uff0c\u4ee5\u53ca\u80fd\u4e0d\u80fd\u7a33\u5b9a\u5730\u8dd1\u5728\u771f\u5b9e\u8bbe\u5907\u4e0a\u3002\u65e0\u8bba\u662f\u63a8\u7406\u6d41\u7a0b\u7684\u7b80\u5316\u3001\u8bad\u7ec3\u7a33\u5b9a\u6027\u7684\u589e\u5f3a\uff0c\u8fd8\u662f\u5bf9 CPU \u548c\u8fb9\u7f18\u573a\u666f\u7684\u9488\u5bf9\u6027\u4f18\u5316\uff0c\u90fd\u6307\u5411\u4e00\u4e2a\u76ee\u6807\uff1a\u8ba9\u89c6\u89c9\u6a21\u578b\u66f4\u5bb9\u6613\u8fdb\u5165\u771f\u5b9e\u7cfb\u7edf\u3002\u5bf9\u4e8e\u9700\u8981\u5728\u4e0d\u540c\u786c\u4ef6\u6761\u4ef6\u4e0b\u6784\u5efa\u89c6\u89c9\u7cfb\u7edf\u7684\u5f00\u53d1\u8005\u6765\u8bf4\uff0c\u8fd9\u79cd\u4ee5\u843d\u5730\u4e3a\u5bfc\u5411\u7684\u6539\u8fdb\uff0c\u63d0\u4f9b\u4e86\u66f4\u591a\u53ef\u64cd\u4f5c\u7684\u7a7a\u95f4\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
          \u539f\u6587\u94fe\u63a5\uff1ahttps:\/\/mp.weixin. […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3,1],"tags":[],"class_list":["post-36582","post","type-post","status-publish","format-standard","hentry","category-technology-frontier","category-home"],"views":0,"_links":{"self":[{"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/posts\/36582"}],"collection":[{"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/comments?post=36582"}],"version-history":[{"count":4,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/posts\/36582\/revisions"}],"predecessor-version":[{"id":36604,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/posts\/36582\/revisions\/36604"}],"wp:attachment":[{"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/media?parent=36582"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/categories?post=36582"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/tags?post=36582"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}