{"id":2367,"date":"2024-06-26T16:17:41","date_gmt":"2024-06-26T08:17:41","guid":{"rendered":"https:\/\/17aitech.com\/?p=2367"},"modified":"2024-10-08T15:18:03","modified_gmt":"2024-10-08T07:18:03","slug":"%e3%80%90%e8%af%be%e7%a8%8b%e6%80%bb%e7%bb%93%e3%80%91day12%ef%bc%9ayolo%e7%9a%84%e6%b7%b1%e5%85%a5%e4%ba%86%e8%a7%a3","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=2367","title":{"rendered":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day12\uff1aYOLO\u7684\u6df1\u5165\u4e86\u89e3"},"content":{"rendered":"<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_78 ez-toc-wrap-left-text counter-hierarchy ez-toc-counter ez-toc-light-blue ez-toc-container-direction\">\n<div class=\"ez-toc-title-container\">\n<p class=\"ez-toc-title\" style=\"cursor:inherit\">\u6587\u7ae0\u76ee\u5f55<\/p>\n<span class=\"ez-toc-title-toggle\"><a href=\"#\" class=\"ez-toc-pull-right ez-toc-btn ez-toc-btn-xs ez-toc-btn-default ez-toc-toggle\" aria-label=\"Toggle Table of Content\"><span class=\"ez-toc-js-icon-con\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #999;color:#999\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #999;color:#999\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/span><\/a><\/span><\/div>\n<nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/17aitech.com\/?p=2367\/#%E5%89%8D%E8%A8%80\" >\u524d\u8a00<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/17aitech.com\/?p=2367\/#YOLO%E9%A1%B9%E7%9B%AE%E7%9A%84%E5%B7%A5%E7%A8%8B%E7%BB%93%E6%9E%84\" >YOLO\u9879\u76ee\u7684\u5de5\u7a0b\u7ed3\u6784<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/17aitech.com\/?p=2367\/#YOLO%E7%9A%84yaml%E6%96%87%E4%BB%B6%E8%A7%A3%E6%9E%90\" >YOLO\u7684yaml\u6587\u4ef6\u89e3\u6790<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/17aitech.com\/?p=2367\/#yaml%E6%96%87%E4%BB%B6%E5%86%85%E5%AE%B9\" >yaml\u6587\u4ef6\u5185\u5bb9<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/17aitech.com\/?p=2367\/#yaml%E6%96%87%E4%BB%B6%E8%A7%A3%E6%9E%90\" >yaml\u6587\u4ef6\u89e3\u6790<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/17aitech.com\/?p=2367\/#YOLOv8_%E6%A8%A1%E5%9E%8B%E7%BB%93%E6%9E%84\" >YOLOv8 \u6a21\u578b\u7ed3\u6784<\/a><\/li><\/ul><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/17aitech.com\/?p=2367\/#YOLO%E7%9A%84%E6%A8%A1%E5%9E%8B%E6%9E%84%E5%BB%BA%E8%BF%87%E7%A8%8B%E8%A7%A3%E6%9E%90\" >YOLO\u7684\u6a21\u578b\u6784\u5efa\u8fc7\u7a0b\u89e3\u6790<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/17aitech.com\/?p=2367\/#%E4%BB%A3%E7%A0%81%E8%B0%83%E7%94%A8%E6%97%B6%E5%BA%8F\" >\u4ee3\u7801\u8c03\u7528\u65f6\u5e8f<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/17aitech.com\/?p=2367\/#%E6%9E%84%E5%BB%BA%E6%A8%A1%E5%9E%8B%E7%9A%84%E6%A0%B8%E5%BF%83%E4%BB%A3%E7%A0%81\" >\u6784\u5efa\u6a21\u578b\u7684\u6838\u5fc3\u4ee3\u7801<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-10\" href=\"https:\/\/17aitech.com\/?p=2367\/#YOLO%E7%9A%84%E6%A8%A1%E5%9E%8B%E8%AE%AD%E7%BB%83%E8%BF%87%E7%A8%8B%E8%A7%A3%E6%9E%90\" >YOLO\u7684\u6a21\u578b\u8bad\u7ec3\u8fc7\u7a0b\u89e3\u6790<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-11\" href=\"https:\/\/17aitech.com\/?p=2367\/#%E4%BB%A3%E7%A0%81%E8%B0%83%E7%94%A8%E6%97%B6%E5%BA%8F-2\" >\u4ee3\u7801\u8c03\u7528\u65f6\u5e8f<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-12\" href=\"https:\/\/17aitech.com\/?p=2367\/#%E6%9E%84%E5%BB%BA%E6%A8%A1%E5%9E%8B%E7%9A%84%E6%A0%B8%E5%BF%83%E4%BB%A3%E7%A0%81-2\" >\u6784\u5efa\u6a21\u578b\u7684\u6838\u5fc3\u4ee3\u7801<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-13\" href=\"https:\/\/17aitech.com\/?p=2367\/#%E5%86%85%E5%AE%B9%E5%B0%8F%E7%BB%93\" >\u5185\u5bb9\u5c0f\u7ed3<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-14\" href=\"https:\/\/17aitech.com\/?p=2367\/#%E5%8F%82%E8%80%83%E8%B5%84%E6%96%99\" >\u53c2\u8003\u8d44\u6599<\/a><\/li><\/ul><\/nav><\/div>\n<h2><span class=\"ez-toc-section\" id=\"%E5%89%8D%E8%A8%80\"><\/span>\u524d\u8a00<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>\u5728<a href=\"https:\/\/17aitech.com\/?p=2322\">\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day11\uff08\u4e0b\uff09\uff1aYOLO\u7684\u5165\u95e8\u4f7f\u7528<\/a>\u4e00\u8282\u4e2d\uff0c\u6211\u4eec\u5df2\u7ecf\u4e86\u89e3YOLO\u7684\u4f7f\u7528\u65b9\u6cd5\uff0c\u4f7f\u7528\u8fc7\u7a0b\u975e\u5e38\u7b80\u5355\uff0c\u8bad\u7ec3\u65f6\u53ea\u9700\u8981\u4e09\u884c\u4ee3\u7801\uff1a\u5f15\u5165YOLO\uff0c\u6784\u5efa\u6a21\u578b\uff0c\u8bad\u7ec3\u6a21\u578b\uff1b\u9884\u6d4b\u65f6\u4e5f\u540c\u6837\u7b80\u5355\uff0c\u53ea\u9700\u8981\u4e24\u884c\u4ee3\u7801\uff1a\u5f15\u5165YOLO\uff0c\u9884\u6d4b\u56fe\u50cf\u5373\u53ef\u3002\u4ee5\u4e0a\u8fc7\u7a0b\u7b80\u5355\u4e3b\u8981\u662fultralytics\u7684\u4ee3\u7801\u5e93\u5df2\u7ecf\u505a\u4e86\u5c01\u88c5\uff0c\u4f7f\u5f97\u4f7f\u7528\u8005\u96c6\u4e2d\u7cbe\u529b\u5728\u6a21\u578b\u8bad\u7ec3\u548c\u9884\u6d4b\u4e0a\u3002<\/p>\n<p>\u4e3a\u4e86\u66f4\u52a0\u6df1\u5165\u4e86\u89e3YOLO\u7684\u5b9e\u73b0\u539f\u7406\uff0c\u672c\u7ae0\u5185\u5bb9\u5c06\u5bf9YOLO\u7684\u5de5\u7a0b\u7ed3\u6784\u3001\u6a21\u578b\u6784\u5efa\u8fc7\u7a0b\u3001\u6a21\u578b\u8bad\u7ec3\u8fc7\u7a0b\u5c1d\u8bd5\u6df1\u5165\u63a2\u7a76\u3002<\/p>\n<h2><span class=\"ez-toc-section\" id=\"YOLO%E9%A1%B9%E7%9B%AE%E7%9A%84%E5%B7%A5%E7%A8%8B%E7%BB%93%E6%9E%84\"><\/span>YOLO\u9879\u76ee\u7684\u5de5\u7a0b\u7ed3\u6784<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<pre><code class=\"language-shell\">|- ultralytics\/\n    |- assets\/    # \u5b58\u653e\u9879\u76ee\u4e2d\u4f7f\u7528\u7684\u8d44\u6e90\u6587\u4ef6\uff0c\u5982\u56fe\u50cf\u3001\u6837\u672c\u6570\u636e\u7b49\u3002\n    |- cfg\/     # \u5b58\u653e\u6a21\u578b\u914d\u7f6e\u6587\u4ef6\uff0c\u5305\u62ec\u4e0d\u540c\u6a21\u578b\u7684\u914d\u7f6e\u4fe1\u606f\uff0c\u5982\u7f51\u7edc\u7ed3\u6784\u3001\u8d85\u53c2\u6570\u7b49\u3002\n    |- data\/    # \u5b58\u653e\u6570\u636e\u96c6\u6587\u4ef6\u548c\u6570\u636e\u5904\u7406\u76f8\u5173\u7684\u4ee3\u7801\u3002\n    |- engine\/  # \u5b58\u653e\u8bad\u7ec3\u548c\u63a8\u7406\u5f15\u64ce\u7684\u4ee3\u7801\uff0c\u5305\u62ec\u8bad\u7ec3\u3001\u6d4b\u8bd5\u3001\u8bc4\u4f30\u7b49\u529f\u80fd\u7684\u5b9e\u73b0\u3002\n    |- hub\/     # \u5b58\u653e\u6a21\u578b\u5e93\u76f8\u5173\u7684\u4ee3\u7801\u548c\u6a21\u578b\u6587\u4ef6\uff0c\u7528\u4e8e\u5feb\u901f\u8c03\u7528\u548c\u4f7f\u7528\u9884\u8bad\u7ec3\u6a21\u578b\u3002\n    |- models\/  # \u5b58\u653e\u6a21\u578b\u7684\u5b9a\u4e49\u548c\u5b9e\u73b0\u4ee3\u7801\uff0c\u5305\u62ec\u4e0d\u540c\u6a21\u578b\u7684\u7f51\u7edc\u7ed3\u6784\u548c\u76f8\u5173\u51fd\u6570\u3002\n    |- nn\/      # \u5b58\u653e\u795e\u7ecf\u7f51\u7edc\u6a21\u5757\u7684\u4ee3\u7801\uff0c\u5305\u62ec\u5404\u79cd\u5c42\u6b21\u7684\u5b9a\u4e49\u548c\u5b9e\u73b0\u3002\n    |- solutions\/ # \u5b58\u653e\u89e3\u51b3\u65b9\u6848\u76f8\u5173\u7684\u4ee3\u7801\u548c\u5b9e\u73b0\uff0c\u7528\u4e8e\u7279\u5b9a\u95ee\u9898\u6216\u4efb\u52a1\u7684\u89e3\u51b3\u65b9\u6848\u3002\n    |- trackers\/  # \u5b58\u653e\u76ee\u6807\u8ddf\u8e2a\u76f8\u5173\u7684\u4ee3\u7801\u548c\u5b9e\u73b0\uff0c\u5305\u62ec\u76ee\u6807\u8ffd\u8e2a\u7b97\u6cd5\u7684\u5b9e\u73b0\u3002\n    |- utils\/   # \u5b58\u653e\u901a\u7528\u7684\u5de5\u5177\u51fd\u6570\u548c\u8f85\u52a9\u51fd\u6570\uff0c\u7528\u4e8e\u9879\u76ee\u4e2d\u7684\u5404\u79cd\u529f\u80fd\u548c\u4efb\u52a1\u3002<\/code><\/pre>\n<p>\u8fdb\u4e00\u6b65\u67e5\u770bcfg\u76ee\u5f55\u7684\u5185\u5bb9\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-shell\">- ultralytics\n  |- cfg\n    |- datasets  # \u6570\u636e\u96c6\u5904\u7406\u548c\u52a0\u8f7d\u76f8\u5173\u6587\u4ef6\n    |- default.yaml  # \u9ed8\u8ba4\u914d\u7f6e\u4fe1\u606f\u6587\u4ef6\n    |- models  # \u5305\u542b\u4e0d\u540c\u6a21\u578b\u7ed3\u6784\u7684\u914d\u7f6e\u6587\u4ef6\n      |- yolov8-cls-resnet101.yaml  # \u5b9a\u4e49 YOLOv8 \u6a21\u578b\u7ed3\u6784\u7684\u914d\u7f6e\u6587\u4ef6\uff08ResNet-101 \u7248\u672c\uff09\n      |- yolov8-cls-resnet50.yaml  # \u5b9a\u4e49 YOLOv8 \u6a21\u578b\u7ed3\u6784\u7684\u914d\u7f6e\u6587\u4ef6\uff08ResNet-50 \u7248\u672c\uff09\n      |- yolov8-cls.yaml  # \u5b9a\u4e49 YOLOv8 \u6a21\u578b\u7ed3\u6784\u7684\u914d\u7f6e\u6587\u4ef6\n      |- ...  # \u5176\u4ed6 YOLOv8 \u6a21\u578b\u7248\u672c\u7684\u914d\u7f6e\u6587\u4ef6\n    |- trackers  # \u76ee\u6807\u8ddf\u8e2a\u76f8\u5173\u6587\u4ef6<\/code><\/pre>\n<h3><span class=\"ez-toc-section\" id=\"YOLO%E7%9A%84yaml%E6%96%87%E4%BB%B6%E8%A7%A3%E6%9E%90\"><\/span>YOLO\u7684yaml\u6587\u4ef6\u89e3\u6790<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<h4><span class=\"ez-toc-section\" id=\"yaml%E6%96%87%E4%BB%B6%E5%86%85%E5%AE%B9\"><\/span>yaml\u6587\u4ef6\u5185\u5bb9<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u4ee5yolov8.yaml\u4e3a\u4f8b\uff0c\u5176\u5185\u5bb9\u5982\u4e0b\uff1a<\/p>\n<pre><code class=\"language-yaml\">nc: 80 # \u7c7b\u522b\u6570\u76ee\uff0cnc\u4ee3\u8868&quot;number of classes&quot;\uff0c\u5373\u6a21\u578b\u7528\u4e8e\u68c0\u6d4b\u7684\u5bf9\u8c61\u7c7b\u522b\u603b\u6570\u3002\nscales: # \u6a21\u578b\u590d\u5408\u7f29\u653e\u5e38\u6570\uff0c\u4f8b\u5982 &#039;model=yolov8n.yaml&#039; \u5c06\u8c03\u7528\u5e26\u6709 &#039;n&#039; \u7f29\u653e\u7684 yolov8.yaml\n  # [depth, width, max_channels]\n  n: [0.33, 0.25, 1024]  # YOLOv8n\u6982\u89c8\uff1a225\u5c42, 3157200\u53c2\u6570, 3157184\u68af\u5ea6, 8.9 GFLOPs\n  s: [0.33, 0.50, 1024]  # YOLOv8s\u6982\u89c8\uff1a225\u5c42, 11166560\u53c2\u6570, 11166544\u68af\u5ea6, 28.8 GFLOPs\n  m: [0.67, 0.75, 768]   # YOLOv8m\u6982\u89c8\uff1a295\u5c42, 25902640\u53c2\u6570, 25902624\u68af\u5ea6, 79.3 GFLOPs\n  l: [1.00, 1.00, 512]   # YOLOv8l\u6982\u89c8\uff1a365\u5c42, 43691520\u53c2\u6570, 43691504\u68af\u5ea6, 165.7 GFLOPs\n  x: [1.00, 1.25, 512]   # YOLOv8x\u6982\u89c8\uff1a365\u5c42, 68229648\u53c2\u6570, 68229632\u68af\u5ea6, 258.5 GFLOPs\n\n# YOLOv8.0n \u9aa8\u5e72\u5c42\nbackbone:\n  # [from, repeats, module, args]\n  - [-1, 1, Conv, [64, 3, 2]]   # 0-P1\/2 \u7b2c0\u5c42\uff0c-1\u4ee3\u8868\u5c06\u4e0a\u5c42\u7684\u8f93\u5165\u4f5c\u4e3a\u672c\u5c42\u7684\u8f93\u5165\u3002\u7b2c0\u5c42\u7684\u8f93\u5165\u662f640*640*3\u7684\u56fe\u50cf\u3002Conv\u4ee3\u8868\u5377\u79ef\u5c42\uff0c\u76f8\u5e94\u7684\u53c2\u6570\uff1a64\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0c3\u4ee3\u8868\u5377\u79ef\u6838\u5927\u5c0fk\uff0c2\u4ee3\u8868stride\u6b65\u957f\u3002\n  - [-1, 1, Conv, [128, 3, 2]]  # 1-P2\/4 \u7b2c1\u5c42\uff0c\u672c\u5c42\u548c\u4e0a\u4e00\u5c42\u662f\u4e00\u6837\u7684\u64cd\u4f5c\uff08128\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0c3\u4ee3\u8868\u5377\u79ef\u6838\u5927\u5c0fk\uff0c2\u4ee3\u8868stride\u6b65\u957f\uff09\n  - [-1, 3, C2f, [128, True]]   #        \u7b2c2\u5c42\uff0c\u672c\u5c42\u662fC2f\u6a21\u5757\uff0c3\u4ee3\u8868\u672c\u5c42\u91cd\u590d3\u6b21\u3002128\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0cTrue\u8868\u793aBottleneck\u6709shortcut\u3002\n  - [-1, 1, Conv, [256, 3, 2]]  # 3-P3\/8 \u7b2c3\u5c42\uff0c\u8fdb\u884c\u5377\u79ef\u64cd\u4f5c\uff08256\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0c3\u4ee3\u8868\u5377\u79ef\u6838\u5927\u5c0fk\uff0c2\u4ee3\u8868stride\u6b65\u957f\uff09\uff0c\u8f93\u51fa\u7279\u5f81\u56fe\u5c3a\u5bf8\u4e3a80*80*256\uff08\u5377\u79ef\u7684\u53c2\u6570\u90fd\u6ca1\u53d8\uff0c\u6240\u4ee5\u90fd\u662f\u957f\u5bbd\u53d8\u6210\u539f\u6765\u76841\/2\uff0c\u548c\u4e4b\u524d\u4e00\u6837\uff09\uff0c\u7279\u5f81\u56fe\u7684\u957f\u5bbd\u5df2\u7ecf\u53d8\u6210\u8f93\u5165\u56fe\u50cf\u76841\/8\u3002\n  - [-1, 6, C2f, [256, True]]   #        \u7b2c4\u5c42\uff0c\u672c\u5c42\u662fC2f\u6a21\u5757\uff0c\u53ef\u4ee5\u53c2\u8003\u7b2c2\u5c42\u7684\u8bb2\u89e3\u30026\u4ee3\u8868\u672c\u5c42\u91cd\u590d6\u6b21\u3002256\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0cTrue\u8868\u793aBottleneck\u6709shortcut\u3002\u7ecf\u8fc7\u8fd9\u5c42\u4e4b\u540e\uff0c\u7279\u5f81\u56fe\u5c3a\u5bf8\u4f9d\u65e7\u662f80*80*256\u3002\n  - [-1, 1, Conv, [512, 3, 2]]  # 5-P4\/16\u7b2c5\u5c42\uff0c\u8fdb\u884c\u5377\u79ef\u64cd\u4f5c\uff08512\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0c3\u4ee3\u8868\u5377\u79ef\u6838\u5927\u5c0fk\uff0c2\u4ee3\u8868stride\u6b65\u957f\uff09\uff0c\u8f93\u51fa\u7279\u5f81\u56fe\u5c3a\u5bf8\u4e3a40*40*512\uff08\u5377\u79ef\u7684\u53c2\u6570\u90fd\u6ca1\u53d8\uff0c\u6240\u4ee5\u90fd\u662f\u957f\u5bbd\u53d8\u6210\u539f\u6765\u76841\/2\uff0c\u548c\u4e4b\u524d\u4e00\u6837\uff09\uff0c\u7279\u5f81\u56fe\u7684\u957f\u5bbd\u5df2\u7ecf\u53d8\u6210\u8f93\u5165\u56fe\u50cf\u76841\/16\u3002\n  - [-1, 6, C2f, [512, True]]   #        \u7b2c6\u5c42\uff0c\u672c\u5c42\u662fC2f\u6a21\u5757\uff0c\u53ef\u4ee5\u53c2\u8003\u7b2c2\u5c42\u7684\u8bb2\u89e3\u30026\u4ee3\u8868\u672c\u5c42\u91cd\u590d6\u6b21\u3002512\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0cTrue\u8868\u793aBottleneck\u6709shortcut\u3002\u7ecf\u8fc7\u8fd9\u5c42\u4e4b\u540e\uff0c\u7279\u5f81\u56fe\u5c3a\u5bf8\u4f9d\u65e7\u662f40*40*512\u3002\n  - [-1, 1, Conv, [1024, 3, 2]] # 7-P5\/32\u7b2c7\u5c42\uff0c\u8fdb\u884c\u5377\u79ef\u64cd\u4f5c\uff081024\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0c3\u4ee3\u8868\u5377\u79ef\u6838\u5927\u5c0fk\uff0c2\u4ee3\u8868stride\u6b65\u957f\uff09\uff0c\u8f93\u51fa\u7279\u5f81\u56fe\u5c3a\u5bf8\u4e3a20*20*1024\uff08\u5377\u79ef\u7684\u53c2\u6570\u90fd\u6ca1\u53d8\uff0c\u6240\u4ee5\u90fd\u662f\u957f\u5bbd\u53d8\u6210\u539f\u6765\u76841\/2\uff0c\u548c\u4e4b\u524d\u4e00\u6837\uff09\uff0c\u7279\u5f81\u56fe\u7684\u957f\u5bbd\u5df2\u7ecf\u53d8\u6210\u8f93\u5165\u56fe\u50cf\u76841\/32\u3002\n  - [-1, 3, C2f, [1024, True]]  #        \u7b2c8\u5c42\uff0c\u672c\u5c42\u662fC2f\u6a21\u5757\uff0c\u53ef\u4ee5\u53c2\u8003\u7b2c2\u5c42\u7684\u8bb2\u89e3\u30023\u4ee3\u8868\u672c\u5c42\u91cd\u590d3\u6b21\u30021024\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0cTrue\u8868\u793aBottleneck\u6709shortcut\u3002\u7ecf\u8fc7\u8fd9\u5c42\u4e4b\u540e\uff0c\u7279\u5f81\u56fe\u5c3a\u5bf8\u4f9d\u65e7\u662f20*20*1024\u3002\n  - [-1, 1, SPPF, [1024, 5]] # 9         \u7b2c9\u5c42\uff0c\u672c\u5c42\u662f\u5feb\u901f\u7a7a\u95f4\u91d1\u5b57\u5854\u6c60\u5316\u5c42\uff08SPPF\uff09\u30021024\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0c5\u4ee3\u8868\u6c60\u5316\u6838\u5927\u5c0fk\u3002\u7ed3\u5408\u6a21\u5757\u7ed3\u6784\u56fe\u548c\u4ee3\u7801\u53ef\u4ee5\u770b\u51fa\uff0c\u6700\u540econcat\u5f97\u5230\u7684\u7279\u5f81\u56fe\u5c3a\u5bf8\u662f20*20*\uff08512*4\uff09\uff0c\u7ecf\u8fc7\u4e00\u6b21Conv\u5f97\u523020*20*1024\u3002\n\n# YOLOv8.0n \u5934\u90e8\u5c42\nhead:\n  - [-1, 1, nn.Upsample, [None, 2, &#039;nearest&#039;]] # \u7b2c10\u5c42\uff0c\u672c\u5c42\u662f\u4e0a\u91c7\u6837\u5c42\u3002-1\u4ee3\u8868\u5c06\u4e0a\u5c42\u7684\u8f93\u51fa\u4f5c\u4e3a\u672c\u5c42\u7684\u8f93\u5165\u3002None\u4ee3\u8868\u4e0a\u91c7\u6837\u7684size\uff08\u8f93\u51fa\u5c3a\u5bf8\uff09\u4e0d\u6307\u5b9a\u30022\u4ee3\u8868scale_factor=2\uff0c\u8868\u793a\u8f93\u51fa\u7684\u5c3a\u5bf8\u662f\u8f93\u5165\u5c3a\u5bf8\u76842\u500d\u3002nearest\u4ee3\u8868\u4f7f\u7528\u7684\u4e0a\u91c7\u6837\u7b97\u6cd5\u4e3a\u6700\u8fd1\u90bb\u63d2\u503c\u7b97\u6cd5\u3002\u7ecf\u8fc7\u8fd9\u5c42\u4e4b\u540e\uff0c\u7279\u5f81\u56fe\u7684\u957f\u548c\u5bbd\u53d8\u6210\u539f\u6765\u7684\u4e24\u500d\uff0c\u901a\u9053\u6570\u4e0d\u53d8\uff0c\u6240\u4ee5\u6700\u7ec8\u5c3a\u5bf8\u4e3a40*40*1024\u3002\n  - [[-1, 6], 1, Concat, [1]]  # cat backbone P4 \u7b2c11\u5c42\uff0c\u672c\u5c42\u662fconcat\u5c42\uff0c[-1, 6]\u4ee3\u8868\u5c06\u4e0a\u5c42\u548c\u7b2c6\u5c42\u7684\u8f93\u51fa\u4f5c\u4e3a\u672c\u5c42\u7684\u8f93\u5165\u3002[1]\u4ee3\u8868concat\u62fc\u63a5\u7684\u7ef4\u5ea6\u662f1\u3002\u4ece\u4e0a\u9762\u7684\u5206\u6790\u53ef\u77e5\uff0c\u4e0a\u5c42\u7684\u8f93\u51fa\u5c3a\u5bf8\u662f40*40*1024\uff0c\u7b2c6\u5c42\u7684\u8f93\u51fa\u662f40*40*512\uff0c\u6700\u7ec8\u672c\u5c42\u7684\u8f93\u51fa\u5c3a\u5bf8\u4e3a40*40*1536\u3002\n  - [-1, 3, C2f, [512]]  # 12 \u7b2c12\u5c42\uff0c\u672c\u5c42\u662fC2f\u6a21\u5757\uff0c\u53ef\u4ee5\u53c2\u8003\u7b2c2\u5c42\u7684\u8bb2\u89e3\u30023\u4ee3\u8868\u672c\u5c42\u91cd\u590d3\u6b21\u3002512\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\u3002\u4e0eBackbone\u4e2dC2f\u4e0d\u540c\u7684\u662f\uff0c\u6b64\u5904\u7684C2f\u7684bottleneck\u6a21\u5757\u7684shortcut=False\u3002\n  - [-1, 1, nn.Upsample, [None, 2, &#039;nearest&#039;]] # \u7b2c13\u5c42\uff0c\u672c\u5c42\u4e5f\u662f\u4e0a\u91c7\u6837\u5c42\uff08\u53c2\u8003\u7b2c10\u5c42\uff09\u3002\u7ecf\u8fc7\u8fd9\u5c42\u4e4b\u540e\uff0c\u7279\u5f81\u56fe\u7684\u957f\u548c\u5bbd\u53d8\u6210\u539f\u6765\u7684\u4e24\u500d\uff0c\u901a\u9053\u6570\u4e0d\u53d8\uff0c\u6240\u4ee5\u6700\u7ec8\u5c3a\u5bf8\u4e3a80*80*512\u3002\n  - [[-1, 4], 1, Concat, [1]]  # cat backbone P3 \u7b2c14\u5c42\uff0c\u672c\u5c42\u662fconcat\u5c42\uff0c[-1, 4]\u4ee3\u8868\u5c06\u4e0a\u5c42\u548c\u7b2c4\u5c42\u7684\u8f93\u51fa\u4f5c\u4e3a\u672c\u5c42\u7684\u8f93\u5165\u3002[1]\u4ee3\u8868concat\u62fc\u63a5\u7684\u7ef4\u5ea6\u662f1\u3002\u4ece\u4e0a\u9762\u7684\u5206\u6790\u53ef\u77e5\uff0c\u4e0a\u5c42\u7684\u8f93\u51fa\u5c3a\u5bf8\u662f80*80*512\uff0c\u7b2c6\u5c42\u7684\u8f93\u51fa\u662f80*80*256\uff0c\u6700\u7ec8\u672c\u5c42\u7684\u8f93\u51fa\u5c3a\u5bf8\u4e3a80*80*768\u3002\n  - [-1, 3, C2f, [256]]  # 15 (P3\/8-small)       \u7b2c15\u5c42\uff0c\u672c\u5c42\u662fC2f\u6a21\u5757\uff0c\u53ef\u4ee5\u53c2\u8003\u7b2c2\u5c42\u7684\u8bb2\u89e3\u30023\u4ee3\u8868\u672c\u5c42\u91cd\u590d3\u6b21\u3002256\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\u3002\u7ecf\u8fc7\u8fd9\u5c42\u4e4b\u540e\uff0c\u7279\u5f81\u56fe\u5c3a\u5bf8\u53d8\u4e3a80*80*256\uff0c\u7279\u5f81\u56fe\u7684\u957f\u5bbd\u5df2\u7ecf\u53d8\u6210\u8f93\u5165\u56fe\u50cf\u76841\/8\u3002\n  - [-1, 1, Conv, [256, 3, 2]] #                 \u7b2c16\u5c42\uff0c\u8fdb\u884c\u5377\u79ef\u64cd\u4f5c\uff08256\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0c3\u4ee3\u8868\u5377\u79ef\u6838\u5927\u5c0fk\uff0c2\u4ee3\u8868stride\u6b65\u957f\uff09\uff0c\u8f93\u51fa\u7279\u5f81\u56fe\u5c3a\u5bf8\u4e3a40*40*256\uff08\u5377\u79ef\u7684\u53c2\u6570\u90fd\u6ca1\u53d8\uff0c\u6240\u4ee5\u90fd\u662f\u957f\u5bbd\u53d8\u6210\u539f\u6765\u76841\/2\uff0c\u548c\u4e4b\u524d\u4e00\u6837\uff09\u3002\n  - [[-1, 12], 1, Concat, [1]]  # cat head P4    \u7b2c17\u5c42\uff0c\u672c\u5c42\u662fconcat\u5c42\uff0c[-1, 12]\u4ee3\u8868\u5c06\u4e0a\u5c42\u548c\u7b2c12\u5c42\u7684\u8f93\u51fa\u4f5c\u4e3a\u672c\u5c42\u7684\u8f93\u5165\u3002[1]\u4ee3\u8868concat\u62fc\u63a5\u7684\u7ef4\u5ea6\u662f1\u3002\u4ece\u4e0a\u9762\u7684\u5206\u6790\u53ef\u77e5\uff0c\u4e0a\u5c42\u7684\u8f93\u51fa\u5c3a\u5bf8\u662f40*40*256\uff0c\u7b2c12\u5c42\u7684\u8f93\u51fa\u662f40*40*512\uff0c\u6700\u7ec8\u672c\u5c42\u7684\u8f93\u51fa\u5c3a\u5bf8\u4e3a40*40*768\u3002\n  - [-1, 3, C2f, [512]]  # 18 (P4\/16-medium)     \u7b2c18\u5c42\uff0c\u672c\u5c42\u662fC2f\u6a21\u5757\uff0c\u53ef\u4ee5\u53c2\u8003\u7b2c2\u5c42\u7684\u8bb2\u89e3\u30023\u4ee3\u8868\u672c\u5c42\u91cd\u590d3\u6b21\u3002512\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\u3002\u7ecf\u8fc7\u8fd9\u5c42\u4e4b\u540e\uff0c\u7279\u5f81\u56fe\u5c3a\u5bf8\u53d8\u4e3a40*40*512\uff0c\u7279\u5f81\u56fe\u7684\u957f\u5bbd\u5df2\u7ecf\u53d8\u6210\u8f93\u5165\u56fe\u50cf\u76841\/16\u3002\n  - [-1, 1, Conv, [512, 3, 2]] #                 \u7b2c19\u5c42\uff0c\u8fdb\u884c\u5377\u79ef\u64cd\u4f5c\uff08512\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\uff0c3\u4ee3\u8868\u5377\u79ef\u6838\u5927\u5c0fk\uff0c2\u4ee3\u8868stride\u6b65\u957f\uff09\uff0c\u8f93\u51fa\u7279\u5f81\u56fe\u5c3a\u5bf8\u4e3a20*20*512\uff08\u5377\u79ef\u7684\u53c2\u6570\u90fd\u6ca1\u53d8\uff0c\u6240\u4ee5\u90fd\u662f\u957f\u5bbd\u53d8\u6210\u539f\u6765\u76841\/2\uff0c\u548c\u4e4b\u524d\u4e00\u6837\uff09\u3002\n  - [[-1, 9], 1, Concat, [1]]  # cat head P5     \u7b2c20\u5c42\uff0c\u672c\u5c42\u662fconcat\u5c42\uff0c[-1, 9]\u4ee3\u8868\u5c06\u4e0a\u5c42\u548c\u7b2c9\u5c42\u7684\u8f93\u51fa\u4f5c\u4e3a\u672c\u5c42\u7684\u8f93\u5165\u3002[1]\u4ee3\u8868concat\u62fc\u63a5\u7684\u7ef4\u5ea6\u662f1\u3002\u4ece\u4e0a\u9762\u7684\u5206\u6790\u53ef\u77e5\uff0c\u4e0a\u5c42\u7684\u8f93\u51fa\u5c3a\u5bf8\u662f20*20*512\uff0c\u7b2c9\u5c42\u7684\u8f93\u51fa\u662f20*20*1024\uff0c\u6700\u7ec8\u672c\u5c42\u7684\u8f93\u51fa\u5c3a\u5bf8\u4e3a20*20*1536\u3002\n  - [-1, 3, C2f, [1024]]  # 21 (P5\/32-large)     \u7b2c21\u5c42\uff0c\u672c\u5c42\u662fC2f\u6a21\u5757\uff0c\u53ef\u4ee5\u53c2\u8003\u7b2c2\u5c42\u7684\u8bb2\u89e3\u30023\u4ee3\u8868\u672c\u5c42\u91cd\u590d3\u6b21\u30021024\u4ee3\u8868\u8f93\u51fa\u901a\u9053\u6570\u3002\u7ecf\u8fc7\u8fd9\u5c42\u4e4b\u540e\uff0c\u7279\u5f81\u56fe\u5c3a\u5bf8\u53d8\u4e3a20*20*1024\uff0c\u7279\u5f81\u56fe\u7684\u957f\u5bbd\u5df2\u7ecf\u53d8\u6210\u8f93\u5165\u56fe\u50cf\u76841\/32\u3002\n  - [[15, 18, 21], 1, Detect, [nc]]  # Detect(P3, P4, P5) \u7b2c20\u5c42\uff0c\u672c\u5c42\u662fDetect\u5c42\uff0c[15, 18, 21]\u4ee3\u8868\u5c06\u7b2c15\u300118\u300121\u5c42\u7684\u8f93\u51fa\uff08\u5206\u522b\u662f80*80*256\u300140*40*512\u300120*20*1024\uff09\u4f5c\u4e3a\u672c\u5c42\u7684\u8f93\u5165\u3002nc\u662f\u6570\u636e\u96c6\u7684\u7c7b\u522b\u6570\u3002\n<\/code><\/pre>\n<h4><span class=\"ez-toc-section\" id=\"yaml%E6%96%87%E4%BB%B6%E8%A7%A3%E6%9E%90\"><\/span>yaml\u6587\u4ef6\u89e3\u6790<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u4e0a\u8ff0\u6587\u4ef6\u5305\u542b\u4e86 YOLOv8 \u6a21\u578b\u7684\u914d\u7f6e\u4fe1\u606f\uff0c\u5176\u4e2d\u5305\u62ec\u4e86\u6a21\u578b\u7684\u7c7b\u522b\u6570\u76ee\u3001\u6a21\u578b\u590d\u5408\u7f29\u653e\u5e38\u6570\u3001\u4e3b\u5e72\u7f51\u7edc\u7ed3\u6784\u548c\u5934\u90e8\u7f51\u7edc\u7ed3\u6784\u3002\u5177\u4f53\u6765\u8bf4\uff1a<\/p>\n<ul>\n<li>nc \u8868\u793a\u6a21\u578b\u7684\u7c7b\u522b\u6570\u76ee\u4e3a 1000\u3002<\/li>\n<li>scales \u5305\u542b\u4e86\u4e0d\u540c\u7f29\u653e\u5e38\u6570\u5bf9\u5e94\u7684\u6a21\u578b\u7ed3\u6784\u53c2\u6570\u3002\n<ul>\n<li>\u5b50\u53c2\u6570\uff1an, s, m, l, x\u8868\u793a\u4e0d\u540c\u7684\u6a21\u578b\u5c3a\u5bf8\uff0c\u6bcf\u4e2a\u5c3a\u5bf8\u90fd\u6709\u5bf9\u5e94\u7684depth\uff08\u6df1\u5ea6\uff09\u3001width\uff08\u5bbd\u5ea6\uff09\u548cmax_channels\uff08\u6700\u5927\u901a\u9053\u6570\uff09\u3002<\/li>\n<li><code>depth<\/code>\uff1a \u8868\u793a\u6df1\u5ea6\u56e0\u5b50\uff0c\u7528\u6765\u63a7\u5236\u4e00\u4e9b\u7279\u5b9a\u6a21\u5757\u7684\u6570\u91cf\u7684\uff0c\u6a21\u5757\u6570\u91cf\u591a\u7f51\u7edc\u6df1\u5ea6\u5c31\u6df1\uff1b<\/li>\n<li><code>width<\/code>\uff1a \u8868\u793a\u5bbd\u5ea6\u56e0\u5b50\uff0c\u7528\u6765\u63a7\u5236\u6574\u4e2a\u7f51\u7edc\u7ed3\u6784\u7684\u901a\u9053\u6570\u91cf\uff0c\u901a\u9053\u6570\u91cf\u8d8a\u591a\uff0c\u7f51\u7edc\u5c31\u770b\u4e0a\u53bb\u66f4\u80d6\u66f4\u5bbd\uff1b<\/li>\n<li><code>max_channels<\/code>\uff1a \u6700\u5927\u901a\u9053\u6570\uff0c\u4e3a\u4e86\u52a8\u6001\u5730\u8c03\u6574\u7f51\u7edc\u7684\u590d\u6742\u6027\u3002\u5728 YOLO \u7684\u65e9\u671f\u7248\u672c\u4e2d\uff0c\u7f51\u7edc\u4e2d\u7684\u6bcf\u4e2a\u5c42\u90fd\u662f\u56fa\u5b9a\u7684\uff0c\u8fd9\u610f\u5473\u7740\u6bcf\u4e2a\u5c42\u7684\u901a\u9053\u6570\u4e5f\u662f\u56fa\u5b9a\u7684\u3002\u4f46\u5728 YOLOv8 \u4e2d\uff0c\u4e3a\u4e86\u589e\u52a0\u7f51\u7edc\u7684\u7075\u6d3b\u6027\u5e76\u4f7f\u5176\u80fd\u591f\u66f4\u597d\u5730\u9002\u5e94\u4e0d\u540c\u7684\u4efb\u52a1\u548c\u6570\u636e\u96c6\uff0c\u5f15\u5165\u4e86 max_channels \u53c2\u6570\u3002<\/li>\n<\/ul>\n<\/li>\n<li>backbone \u5b9a\u4e49\u4e86 YOLOv8.0n \u6a21\u578b\u7684\u4e3b\u5e72\u7f51\u7edc\u7ed3\u6784\uff0c\u5305\u62ec\u4e86\u5377\u79ef\u5c42\u3001C2f \u6a21\u5757\u7b49\u3002\n<ul>\n<li><code>from(\u6765\u81ea)<\/code>\uff1a\n<ul>\n<li>\u8fd9\u4e2a\u5b57\u6bb5\u8868\u793a\u5f53\u524d\u5c42\u8fde\u63a5\u5230\u7684\u4e0a\u4e00\u5c42\u7684\u7d22\u5f15\u3002\u901a\u5e38\uff0c-1 \u8868\u793a\u8fde\u63a5\u5230\u4e0a\u4e00\u5c42\uff0c0 \u8868\u793a\u8fde\u63a5\u5230\u8f93\u5165\u6570\u636e\u3002<\/li>\n<li>\u4f8b\u5982\uff0c[-1, 1, Conv, [64, 3, 2]] \u8868\u793a\u5f53\u524d\u5c42\u8fde\u63a5\u5230\u4e0a\u4e00\u5c42\uff0c\u5373\u524d\u4e00\u5c42\u7684\u8f93\u51fa\u4f5c\u4e3a\u5f53\u524d\u5c42\u7684\u8f93\u5165\u3002<\/li>\n<\/ul>\n<\/li>\n<li><code>repeats(\u91cd\u590d\u6b21\u6570)<\/code>\uff1a\n<ul>\n<li>\u8fd9\u4e2a\u5b57\u6bb5\u8868\u793a\u5f53\u524d\u5c42\u7684\u6a21\u5757\uff08module\uff09\u88ab\u91cd\u590d\u4f7f\u7528\u7684\u6b21\u6570\u3002<\/li>\n<li>\u4f8b\u5982\uff0c[-1, 3, C2f, [128, True]] \u8868\u793a\u5f53\u524d\u5c42\u7684\u6a21\u5757 C2f \u4f1a\u88ab\u91cd\u590d\u4f7f\u7528 3 \u6b21\u3002<\/li>\n<\/ul>\n<\/li>\n<li><code>module(\u6a21\u5757)<\/code>\uff1a\n<ul>\n<li>\u8fd9\u4e2a\u5b57\u6bb5\u8868\u793a\u5f53\u524d\u5c42\u4f7f\u7528\u7684\u6a21\u5757\u7c7b\u578b\uff0c\u5982 Conv\uff08\u5377\u79ef\u5c42\uff09\u3001C2f \u7b49\u3002<\/li>\n<li>\u4f8b\u5982\uff0c[-1, 1, Conv, [64, 3, 2]] \u4e2d\u7684 Conv \u8868\u793a\u5f53\u524d\u5c42\u4f7f\u7528\u7684\u662f\u5377\u79ef\u5c42\u3002<\/li>\n<\/ul>\n<\/li>\n<li><code>args(\u53c2\u6570)<\/code>\uff1a\n<ul>\n<li>\u8fd9\u4e2a\u5b57\u6bb5\u5305\u542b\u4e86\u5f53\u524d\u5c42\u6a21\u5757\u7684\u53c2\u6570\uff0c\u4f8b\u5982\u5377\u79ef\u5c42\u7684\u901a\u9053\u6570\u3001\u5377\u79ef\u6838\u5927\u5c0f\u3001\u6b65\u957f\u7b49\u3002<\/li>\n<li>\u4f8b\u5982\uff0c[-1, 1, Conv, [64, 3, 2]] \u4e2d\u7684 [64, 3, 2] \u8868\u793a\u5377\u79ef\u5c42\u7684\u901a\u9053\u6570\u4e3a 64\uff0c\u5377\u79ef\u6838\u5927\u5c0f\u4e3a 3\uff0c\u6b65\u957f\u4e3a 2\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<h4><span class=\"ez-toc-section\" id=\"YOLOv8_%E6%A8%A1%E5%9E%8B%E7%BB%93%E6%9E%84\"><\/span>YOLOv8 \u6a21\u578b\u7ed3\u6784<span class=\"ez-toc-section-end\"><\/span><\/h4>\n<p>\u4e0a\u8ff0yaml\u5b9a\u4e49\u7684\u6a21\u578b\u7ed3\u6784\uff0c\u4f7f\u7528\u56fe\u793a\u663e\u793a\u5982\u4e0b\uff1a<br \/>\n<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLOV8\u6a21\u578b\u7ed3\u6784\u89e3\u6790.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLOV8\u6a21\u578b\u7ed3\u6784\u89e3\u6790.png\" alt=\"\" \/><\/a><br \/>\n\u4e3b\u8981\u7ec4\u6210\u90e8\u5206\uff1a<\/p>\n<ul>\n<li>Backbone\uff08\u4e3b\u5e72\u7f51\u7edc\uff09\n<pre><code>\u4e3b\u5e72\u7f51\u7edc\u662f\u6a21\u578b\u7684\u57fa\u7840\uff0c\u8d1f\u8d23\u4ece\u8f93\u5165\u56fe\u50cf\u4e2d\u63d0\u53d6\u7279\u5f81\u3002\u8fd9\u4e9b\u7279\u5f81\u662f\u540e\u7eed\u7f51\u7edc\u5c42\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\u7684\u57fa\u7840\u3002<\/code><\/pre>\n<\/li>\n<li>Head\uff08\u5934\u90e8\u7f51\u7edc\uff09\n<pre><code>\u5934\u90e8\u7f51\u7edc\u662f\u76ee\u6807\u68c0\u6d4b\u6a21\u578b\u7684\u51b3\u7b56\u90e8\u5206\uff0c\u8d1f\u8d23\u4ea7\u751f\u6700\u7ec8\u7684\u68c0\u6d4b\u7ed3\u679c\u3002<\/code><\/pre>\n<\/li>\n<li>ConvModule\n<pre><code>\u5305\u542b\u5377\u79ef\u5c42\u3001BN\uff08\u6279\u91cf\u5f52\u4e00\u5316\uff09\u548c\u6fc0\u6d3b\u51fd\u6570\uff08\u5982SiLU\uff09\uff0c\u7528\u4e8e\u63d0\u53d6\u7279\u5f81\u3002<\/code><\/pre>\n<\/li>\n<li>DarknetBottleneck\uff1a\n<pre><code>\u901a\u8fc7residual connections\uff08\u6b8b\u5dee\u7ed3\u6784\uff09\u589e\u52a0\u7f51\u7edc\u6df1\u5ea6\uff0c\u540c\u65f6\u4fdd\u6301\u6548\u7387\u3002<\/code><\/pre>\n<\/li>\n<li>CSP Layer\uff1a\n<pre><code>CSP\u7ed3\u6784\u7684\u53d8\u4f53\uff0c\u901a\u8fc7\u90e8\u5206\u8fde\u63a5\u6765\u63d0\u9ad8\u6a21\u578b\u7684\u8bad\u7ec3\u6548\u7387\u3002<\/code><\/pre>\n<\/li>\n<li>Concat\uff1a\n<pre><code>\u7279\u5f81\u56fe\u62fc\u63a5\uff0c\u7528\u4e8e\u5408\u5e76\u4e0d\u540c\u5c42\u7684\u7279\u5f81\u3002<\/code><\/pre>\n<\/li>\n<li>Upsample\uff1a\n<pre><code>\u4e0a\u91c7\u6837\u64cd\u4f5c\uff0c\u589e\u52a0\u7279\u5f81\u56fe\u7684\u7a7a\u95f4\u5206\u8fa8\u7387\u3002<\/code><\/pre>\n<\/li>\n<\/ul>\n<p>\u4e3b\u8981\u8fc7\u7a0b\u4e3a\uff1a<br \/>\n1.<strong>\u591a\u5c42\u5377\u79ef<\/strong>\uff1a\u56fe\u7247\u8f93\u5165\u5230\u4e3b\u5e72\u7f51\u7edc\uff0c\u7ecf\u8fc7P1-P5\u5c42\u5377\u79ef\u540e\uff0c\u5728\u7b2c9\u5c42\u901a\u8fc7SPPF\u6a21\u5757\u8fdb\u884c\u7279\u5f81\u63d0\u53d6\u3002<br \/>\n2.<strong>\u4e0a\u91c7\u6837\u548c\u8fde\u63a5<\/strong>\uff1a\u7ecf\u8fc7SPPF\u540e\u8fdb\u884cUpsample\u4e0a\u91c7\u6837\u64cd\u4f5c\uff0c\u7136\u540e\u4e0e\u7b2c6\u5c42\u8fdb\u884cconcat\u64cd\u4f5c\uff0c\u518d\u8fdb\u884cC2f\u6a21\u5757\u7279\u5f81\u63d0\u53d6\u3002<br \/>\n\u7c7b\u4f3c\u7684:<\/p>\n<ul>\n<li>\u572814\u5c42\u4e0e\u7b2c4\u5c42\u8fdb\u884cconcat\u64cd\u4f5c\uff0c\u518d\u8fdb\u884cC2f\u6a21\u5757\u7279\u5f81\u63d0\u53d6\uff1b<\/li>\n<li>\u572820\u5c42\u4e0e\u7b2c9\u5c42\u8fdb\u884cconcat\u64cd\u4f5c\uff0c\u518d\u8fdb\u884cC2f\u6a21\u5757\u7279\u5f81\u63d0\u53d6\u3002<\/li>\n<\/ul>\n<p>3.<strong>\u76ee\u6807\u68c0\u6d4b<\/strong>\uff1a\u6700\u540e\u5c0615\u5c42\u300118\u5c42\u300121\u5c42\u5206\u522b\u7ecf\u8fc7Detect\u6a21\u5757\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\u3002<\/p>\n<h2><span class=\"ez-toc-section\" id=\"YOLO%E7%9A%84%E6%A8%A1%E5%9E%8B%E6%9E%84%E5%BB%BA%E8%BF%87%E7%A8%8B%E8%A7%A3%E6%9E%90\"><\/span>YOLO\u7684\u6a21\u578b\u6784\u5efa\u8fc7\u7a0b\u89e3\u6790<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<h3><span class=\"ez-toc-section\" id=\"%E4%BB%A3%E7%A0%81%E8%B0%83%E7%94%A8%E6%97%B6%E5%BA%8F\"><\/span>\u4ee3\u7801\u8c03\u7528\u65f6\u5e8f<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLOV8\u65f6\u5e8f\u56fe\u2014\u6784\u9020\u6a21\u578b.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLOV8\u65f6\u5e8f\u56fe\u2014\u6784\u9020\u6a21\u578b.png\" alt=\"\" \/><\/a><\/p>\n<h3><span class=\"ez-toc-section\" id=\"%E6%9E%84%E5%BB%BA%E6%A8%A1%E5%9E%8B%E7%9A%84%E6%A0%B8%E5%BF%83%E4%BB%A3%E7%A0%81\"><\/span>\u6784\u5efa\u6a21\u578b\u7684\u6838\u5fc3\u4ee3\u7801<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p>\u901a\u8fc7\u67e5\u770b\u4e0a\u8ff0YOLO\u6784\u5efa\u6a21\u578b\u8fc7\u7a0b\uff0c\u5176\u6838\u5fc3\u4ee3\u7801\u4e3b\u8981\u662f\u4ee5\u4e0b\u4e24\u5904\uff1a<br \/>\n\u6838\u5fc3\u4ee3\u78011\uff1aself._smart_load()<\/p>\n<pre><code class=\"language-python\">    def _new(self, cfg: str, task=None, model=None, verbose=False) -&gt; None:\n        # \u4ee5\u4e0a\u90e8\u5206\u7701\u7565...\n        self.model = (model or self._smart_load(&quot;model&quot;))(cfg_dict, verbose=verbose and RANK == -1)  # \u6838\u5fc3\u4ee3\u7801\n        self.overrides[&quot;model&quot;] = self.cfg\n        self.overrides[&quot;task&quot;] = self.task\n\n        self.model.args = {**DEFAULT_CFG_DICT, **self.overrides}  # \u6838\u5fc3\u4ee3\u7801\n        # \u4ee5\u4e0b\u90e8\u5206\u7701\u7565...<\/code><\/pre>\n<p>\u4ee3\u7801\u89e3\u6790\uff1a<br \/>\n<code>self.model = (model or self._smart_load(&quot;model&quot;))(cfg_dict, verbose=verbose and RANK == -1)<\/code><\/p>\n<ul>\n<li>\u7279\u6027\uff1a\u4f7f\u7528 Python \u7684\u6761\u4ef6\u8868\u8fbe\u5f0f\uff0c\u9009\u62e9\u4e0d\u540c\u7684\u51fd\u6570\u6216\u5bf9\u8c61\u8fdb\u884c\u8d4b\u503c<\/li>\n<li>\u89e3\u91ca\u8bf4\u660e\uff1a\n<ul>\n<li><code>model or self._smart_load(&quot;model&quot;)<\/code> \u662f\u4e00\u4e2a\u6761\u4ef6\u8868\u8fbe\u5f0f\uff0c\u5b83\u4f1a\u6839\u636e <code>model<\/code> \u53d8\u91cf\u662f\u5426\u4e3a\u771f\u503c\u6765\u9009\u62e9\u8d4b\u503c\u7684\u5bf9\u8c61\u3002<\/li>\n<li>\u5982\u679c <code>model<\/code> \u4e3a\u771f\u503c\uff08\u975e\u7a7a\uff09\uff0c\u5219 <code>self.model<\/code> \u5c06\u88ab\u8d4b\u503c\u4e3a <code>model<\/code>\u3002<\/li>\n<li>\u5982\u679c <code>model<\/code> \u4e3a\u5047\u503c\uff08\u7a7a\uff09\uff0c\u5219\u4f1a\u8c03\u7528 <code>self._smart_load(&quot;model&quot;)<\/code> \u65b9\u6cd5\u6765\u52a0\u8f7d\u6a21\u578b\uff0c\u5e76\u5c06\u8fd4\u56de\u7684\u5bf9\u8c61\u8d4b\u503c\u7ed9 <code>self.model<\/code>\u3002<\/li>\n<li>\u7136\u540e\u4e0e\uff0c\u7b2c\u4e8c\u4e2a\u62ec\u53f7<code>(cfg_dict, verbose=verbose and RANK == -1)<\/code> \u62fc\u63a5\uff0c\u5f62\u6210<code>self.model(cfg_dict, verbose=verbose and RANK == -1)<\/code><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p><code>self.model.args = {**DEFAULT_CFG_DICT, **self.overrides}<\/code><\/p>\n<ul>\n<li>\u7279\u6027\uff1a\u4f7f\u7528 Python \u4e2d\u7684\u5b57\u5178\u5408\u5e76\u64cd\u4f5c\u7b26 ** \u6765\u5c06\u4e24\u4e2a\u5b57\u5178\u5408\u5e76\u3002<\/li>\n<li>\u89e3\u91ca\u8bf4\u660e\uff1a\n<ul>\n<li><code>DEFAULT_CFG_DICT<\/code> \u548c <code>self.overrides<\/code> \u662f\u4e24\u4e2a\u5b57\u5178\u3002<\/li>\n<li><code>**DEFAULT_CFG_DICT<\/code> \u5c06 <code>DEFAULT_CFG_DICT<\/code> \u5b57\u5178\u4e2d\u7684\u6240\u6709\u952e\u503c\u5bf9\u89e3\u5305\u5e76\u6dfb\u52a0\u5230\u65b0\u7684\u5b57\u5178\u4e2d\u3002<\/li>\n<li><code>**self.overrides<\/code> \u540c\u6837\u5c06 <code>self.overrides<\/code> \u5b57\u5178\u4e2d\u7684\u6240\u6709\u952e\u503c\u5bf9\u89e3\u5305\u5e76\u6dfb\u52a0\u5230\u540c\u4e00\u4e2a\u65b0\u7684\u5b57\u5178\u4e2d\u3002<\/li>\n<li>\u6700\u7ec8\uff0c<code>{\\*\\*DEFAULT_CFG_DICT, \\*\\*self.overrides} <\/code>\u8868\u793a\u5c06\u8fd9\u4e24\u4e2a\u5b57\u5178\u5408\u5e76\u6210\u4e00\u4e2a\u65b0\u7684\u5b57\u5178\uff0c\u5176\u4e2d <code>self.overrides<\/code>\u4e2d\u7684\u952e\u503c\u5bf9\u5c06\u8986\u76d6 <code>DEFAULT_CFG_DICT<\/code> \u4e2d\u7684\u540c\u540d\u952e\u503c\u5bf9\u3002<\/li>\n<\/ul>\n<\/li>\n<li>\u4e3e\u4f8b\uff1a<\/li>\n<\/ul>\n<pre><code class=\"language-python\">dict1 = {&#039;a&#039;: 1, &#039;b&#039;: 2}\ndict2 = {&#039;b&#039;: 3, &#039;c&#039;: 4}\nmerged_dict = {**dict1, **dict2}\nprint(merged_dict)\n# \u8fd0\u884c\u7ed3\u679c\uff1a\n# {&#039;a&#039;: 1, &#039;b&#039;: 3, &#039;c&#039;: 4}<\/code><\/pre>\n<p>\u6838\u5fc3\u4ee3\u78012\uff1aself.task_map[self.task][key]<\/p>\n<pre><code class=\"language-python\">def task_map(self):\n        &quot;&quot;&quot;Map head to model, trainer, validator, and predictor classes.&quot;&quot;&quot;\n        return {\n            &quot;classify&quot;: {\n                &quot;model&quot;: ClassificationModel,\n                &quot;trainer&quot;: yolo.classify.ClassificationTrainer,\n                &quot;validator&quot;: yolo.classify.ClassificationValidator,\n                &quot;predictor&quot;: yolo.classify.ClassificationPredictor,\n            },\n            &quot;detect&quot;: {\n                &quot;model&quot;: DetectionModel,\n                &quot;trainer&quot;: yolo.detect.DetectionTrainer,\n                &quot;validator&quot;: yolo.detect.DetectionValidator,\n                &quot;predictor&quot;: yolo.detect.DetectionPredictor,\n            },\n            # \u4ee5\u4e0b\u90e8\u5206\u7701\u7565...\n        }<\/code><\/pre>\n<p>\u4ee3\u7801\u89e3\u6790\uff1a<\/p>\n<ul>\n<li>\u4ee5\u4e0a\u4ee3\u7801\u6839\u636e\u4f20\u5165\u7684self.task\u7c7b\u578b(\u4f8b\u5982\uff1aclassify)\uff0c\u6765\u521b\u5efa\u5bf9\u5e94\u7684\u6a21\u578b\u3001\u8bad\u7ec3\u5668\u3001\u9a8c\u8bc1\u5668\u3001\u9884\u6d4b\u5668\u7c7b\u3002<\/li>\n<li>\u7279\u6027\uff1a\u4f7f\u7528\u4e86 Python \u4e2d\u7684\u5b57\u5178\uff0c\u5c06\u5b57\u7b26\u4e32\u952e\u4e0e\u7c7b\u5bf9\u8c61\u503c\u8fdb\u884c\u5173\u8054\uff0c\u4ee5\u5b9e\u73b0\u5c06\u7c7b\u5bf9\u8c61\u6620\u5c04\u5230\u4e0d\u540c\u7684\u529f\u80fd\u6a21\u5757<\/li>\n<li>\u5b57\u5178\u4e0e\u7c7b\u5bf9\u8c61\u6620\u5c04\u7684\u4e3e\u4f8b<\/li>\n<\/ul>\n<pre><code class=\"language-python\">class Dog:\n    def __init__(self, name):\n        self.name = name\nclass Cat:\n    def __init__(self, name):\n        self.name = name\n# \u521b\u5efa\u4e00\u4e2a\u5b57\u5178\uff0c\u5c06\u5b57\u7b26\u4e32\u952e\u6620\u5c04\u5230\u4e0d\u540c\u7684\u7c7b\u5bf9\u8c61\nanimal_map = {\n    &quot;dog&quot;: Dog,\n    &quot;cat&quot;: Cat,\n}\n# \u6839\u636e\u952e\u6765\u5b9e\u4f8b\u5316\u4e0d\u540c\u7684\u7c7b\u5bf9\u8c61\nmy_dog = animal_map[&quot;dog&quot;](&quot;Buddy&quot;)\nmy_cat = animaljson_map[&quot;cat&quot;](&quot;Whiskers&quot;)\nprint(my_dog.name)  # \u8f93\u51fa: Buddy\nprint(my_cat.name)  # \u8f93\u51fa: Whiskers\n<\/code><\/pre>\n<p>\u6838\u5fc3\u4ee3\u78013\uff1aparse_model()\u51fd\u6570<\/p>\n<pre><code class=\"language-python\">def parse_model(d, ch, verbose=True):  # model_dict, input_channels(3)\n    &quot;&quot;&quot;Parse a YOLO model.yaml dictionary into a PyTorch model.&quot;&quot;&quot;\n    import ast\n\n    # Args\n    max_channels = float(&quot;inf&quot;)\n    nc, act, scales = (d.get(x) for x in (&quot;nc&quot;, &quot;activation&quot;, &quot;scales&quot;))\n    depth, width, kpt_shape = (d.get(x, 1.0) for x in (&quot;depth_multiple&quot;, &quot;width_multiple&quot;, &quot;kpt_shape&quot;))\n    if scales:\n        scale = d.get(&quot;scale&quot;)\n        if not scale:\n            scale = tuple(scales.keys())[0]\n            LOGGER.warning(f&quot;WARNING \u26a0\ufe0f no model scale passed. Assuming scale=&#039;{scale}&#039;.&quot;)\n        depth, width, max_channels = scales[scale]\n\n    if act:\n        Conv.default_act = eval(act)  # redefine default activation, i.e. Conv.default_act = nn.SiLU()\n        if verbose:\n            LOGGER.info(f&quot;{colorstr(&#039;activation:&#039;)} {act}&quot;)  # print\n\n    ch = [ch]\n    layers, save, c2 = [], [], ch[-1]  # layers, savelist, ch out\n    for i, (f, n, m, args) in enumerate(d[&quot;backbone&quot;] + d[&quot;head&quot;]):  # from, number, module, args\n        m = getattr(torch.nn, m[3:]) if &quot;nn.&quot; in m else globals()[m]  # get module\n        for j, a in enumerate(args):\n            if isinstance(a, str):\n                with contextlib.suppress(ValueError):\n                    args[j] = locals()[a] if a in locals() else ast.literal_eval(a)\n\n        n = n_ = max(round(n * depth), 1) if n &gt; 1 else n  # depth gain\n        if m in {\n            Classify,\n            Conv,\n            ConvTranspose,\n            GhostConv,\n            Bottleneck,\n            GhostBottleneck,\n            SPP,\n            SPPF,\n            DWConv,\n            Focus,\n            BottleneckCSP,\n            C1,\n            C2,\n            C2f,\n            RepNCSPELAN4,\n            ELAN1,\n            ADown,\n            AConv,\n            SPPELAN,\n            C2fAttn,\n            C3,\n            C3TR,\n            C3Ghost,\n            nn.ConvTranspose2d,\n            DWConvTranspose2d,\n            C3x,\n            RepC3,\n            PSA,\n            SCDown,\n            C2fCIB,\n        }:\n            c1, c2 = ch[f], args[0]\n            if c2 != nc:  # if c2 not equal to number of classes (i.e. for Classify() output)\n                c2 = make_divisible(min(c2, max_channels) * width, 8)\n            if m is C2fAttn:\n                args[1] = make_divisible(min(args[1], max_channels \/\/ 2) * width, 8)  # embed channels\n                args[2] = int(\n                    max(round(min(args[2], max_channels \/\/ 2 \/\/ 32)) * width, 1) if args[2] &gt; 1 else args[2]\n                )  # num heads\n\n            args = [c1, c2, *args[1:]]\n        # ...((\u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u505a\u7701\u7565))\n\n        m_ = nn.Sequential(*(m(*args) for _ in range(n))) if n &gt; 1 else m(*args)  # module\n        t = str(m)[8:-2].replace(&quot;__main__.&quot;, &quot;&quot;)  # module type\n        m.np = sum(x.numel() for x in m_.parameters())  # number params\n        m_.i, m_.f, m_.type = i, f, t  # attach index, &#039;from&#039; index, type\n        if verbose:\n            LOGGER.info(f&quot;{i:&gt;3}{str(f):&gt;20}{n_:&gt;3}{m.np:10.0f}  {t:&lt;45}{str(args):&lt;30}&quot;)  # print\n        save.extend(x % i for x in ([f] if isinstance(f, int) else f) if x != -1)  # append to savelist\n        layers.append(m_)\n        if i == 0:\n            ch = []\n        ch.append(c2)\n    return nn.Sequential(*layers), sorted(save)<\/code><\/pre>\n<p>\u4ee3\u7801\u89e3\u6790\uff1a<\/p>\n<ul>\n<li>\u53c2\u6570\u8bbe\u7f6e\uff1a\n<ul>\n<li>\u4ee5\u4e0a\u4ee3\u7801\u5b9a\u4e49\u4e86\u4e00\u4e9b\u521d\u59cb\u53c2\u6570\uff0c\u5982 max_channels\u3001nc\uff08\u7c7b\u522b\u6570\uff09\u3001act\uff08\u6fc0\u6d3b\u51fd\u6570\u7c7b\u578b\uff09\u3001scales\uff08\u6a21\u578b\u5c3a\u5ea6\uff09\u3001depth\uff08\u6df1\u5ea6\u500d\u6570\uff09\u3001width\uff08\u5bbd\u5ea6\u500d\u6570\uff09\u7b49\u3002<\/li>\n<li>\u6839\u636e\u914d\u7f6e\u6587\u4ef6\u4e2d\u7684 scales \u53c2\u6570\u8bbe\u7f6e\u6a21\u578b\u7684\u6df1\u5ea6\u3001\u5bbd\u5ea6\u548c\u6700\u5927\u901a\u9053\u6570\u3002<\/li>\n<\/ul>\n<\/li>\n<li>\u6fc0\u6d3b\u51fd\u6570\u8bbe\u7f6e\uff1a\n<ul>\n<li>\u5982\u679c\u914d\u7f6e\u6587\u4ef6\u4e2d\u6307\u5b9a\u4e86\u6fc0\u6d3b\u51fd\u6570\u7c7b\u578b act\uff0c\u5219\u91cd\u65b0\u5b9a\u4e49\u9ed8\u8ba4\u6fc0\u6d3b\u51fd\u6570\u4e3a\u6307\u5b9a\u7684\u6fc0\u6d3b\u51fd\u6570\uff08\u5982 nn.SiLU()\uff09\u3002<\/li>\n<\/ul>\n<\/li>\n<li>\u6a21\u578b\u89e3\u6790\uff1a\n<ul>\n<li>\u904d\u5386\u914d\u7f6e\u6587\u4ef6\u4e2d\u7684 backbone \u548c head \u90e8\u5206\uff0c\u89e3\u6790\u6bcf\u4e2a\u5c42\u7684\u4fe1\u606f\u3002<\/li>\n<li>\u6839\u636e\u6a21\u5757\u7c7b\u578b m\uff08\u5982 Conv\u3001Bottleneck \u7b49\uff09\u9009\u62e9\u76f8\u5e94\u7684\u5904\u7406\u903b\u8f91\uff0c\u8bbe\u7f6e\u8f93\u5165\u901a\u9053 c1\u3001\u8f93\u51fa\u901a\u9053 c2 \u4ee5\u53ca\u5176\u4ed6\u53c2\u6570\u3002<\/li>\n<\/ul>\n<\/li>\n<li>\u521b\u5efa\u6a21\u5757\uff1a\n<ul>\n<li>\u6839\u636e\u89e3\u6790\u5f97\u5230\u7684\u53c2\u6570\u548c\u6a21\u5757\u7c7b\u578b\uff0c\u901a\u8fc7<code>nn.Sequential(*layers)<\/code>\u521b\u5efa\u76f8\u5e94\u7684\u5c42\uff0c\u6700\u7ec8\u8fd4\u56de\u89e3\u6790\u540e\u7684\u6a21\u578b\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"YOLO%E7%9A%84%E6%A8%A1%E5%9E%8B%E8%AE%AD%E7%BB%83%E8%BF%87%E7%A8%8B%E8%A7%A3%E6%9E%90\"><\/span>YOLO\u7684\u6a21\u578b\u8bad\u7ec3\u8fc7\u7a0b\u89e3\u6790<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<h3><span class=\"ez-toc-section\" id=\"%E4%BB%A3%E7%A0%81%E8%B0%83%E7%94%A8%E6%97%B6%E5%BA%8F-2\"><\/span>\u4ee3\u7801\u8c03\u7528\u65f6\u5e8f<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLOV8\u65f6\u5e8f\u56fe\u2014\u8bad\u7ec3\u6a21\u578b.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLOV8\u65f6\u5e8f\u56fe\u2014\u8bad\u7ec3\u6a21\u578b.png\" alt=\"\" \/><\/a><\/p>\n<h3><span class=\"ez-toc-section\" id=\"%E6%9E%84%E5%BB%BA%E6%A8%A1%E5%9E%8B%E7%9A%84%E6%A0%B8%E5%BF%83%E4%BB%A3%E7%A0%81-2\"><\/span>\u6784\u5efa\u6a21\u578b\u7684\u6838\u5fc3\u4ee3\u7801<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p>\u6838\u5fc3\u4ee3\u78011\uff1aClassificationDataset\u6570\u636e\u96c6<\/p>\n<pre><code class=\"language-python\">\n    def __init__(self, root, args, augment=False, prefix=&quot;&quot;):\n        # \u4ee5\u4e0a\u5185\u5bb9\u7701\u7565\n        self.torch_transforms = (\n            classify_augmentations(\n                size=args.imgsz,\n                scale=scale,\n                hflip=args.fliplr,\n                vflip=args.flipud,\n                erasing=args.erasing,\n                auto_augment=args.auto_augment,\n                hsv_h=args.hsv_h,\n                hsv_s=args.hsv_s,\n                hsv_v=args.hsv_v,\n            )\n            if augment\n            else classify_transforms(size=args.imgsz, crop_fraction=args.crop_fraction)\n        )<\/code><\/pre>\n<p>\u4ee3\u7801\u89e3\u6790\uff1a<\/p>\n<ul>\n<li>\u4ee5\u4e0a\u4ee3\u7801\u662fClassificationDataset\u6570\u636e\u96c6\u7684\u521d\u59cb\u5316\u51fd\u6570<\/li>\n<li>self.torch_transforms = &#8230;\uff1a\n<ul>\n<li>\u6839\u636e\u6761\u4ef6\u9009\u62e9\u4e0d\u540c\u7684\u6570\u636e\u589e\u5f3a\u65b9\u6cd5\uff1a<\/li>\n<li>\u5982\u679c augment \u4e3a True\uff0c\u5219\u8c03\u7528 classify_augmentations() \u65b9\u6cd5\u8fdb\u884c\u6570\u636e\u589e\u5f3a\u3002<\/li>\n<li>\u5426\u5219\uff0c\u8c03\u7528 classify_transforms() \u65b9\u6cd5\u8fdb\u884c\u6570\u636e\u8f6c\u6362\u3002<\/li>\n<\/ul>\n<\/li>\n<li>\u9664\u6b64\u4e4b\u5916\uff0c\u8be5\u6570\u636e\u96c6\u6309\u7167\u6807\u51c6\u89c4\u8303\uff0c\u5b9e\u73b0\u4e86<strong>getitem<\/strong>(),<strong>len<\/strong>()\u7b49\u56de\u8c03\u51fd\u6570\u3002<\/li>\n<\/ul>\n<p>\u6838\u5fc3\u4ee3\u78012\uff1a\u7b79\u5907\u8bad\u7ec3<\/p>\n<pre><code class=\"language-python\">    def _setup_train(self, world_size):\n        &quot;&quot;&quot;Builds dataloaders and optimizer on correct rank process.&quot;&quot;&quot;\n\n        # Model\n        self.run_callbacks(&quot;on_pretrain_routine_start&quot;)\n        ckpt = self.setup_model()\n        self.model = self.model.to(self.device)\n        self.set_model_attributes()\n\n        # Freeze layers\n        # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n        # Check AMP\n        # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n        # Check imgsz\n        gs = max(int(self.model.stride.max() if hasattr(self.model, &quot;stride&quot;) else 32), 32)  # grid size (max stride)\n        self.args.imgsz = check_imgsz(self.args.imgsz, stride=gs, floor=gs, max_dim=1)\n        self.stride = gs  # for multiscale training\n\n        # Batch size\n        # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n        # Dataloaders\n        batch_size = self.batch_size \/\/ max(world_size, 1)\n        self.train_loader = self.get_dataloader(self.trainset, batch_size=batch_size, rank=RANK, mode=&quot;train&quot;)\n        # \u4ee3\u7801\u5df2\u7701\u7565\n\n        # Optimizer\n        self.accumulate = max(round(self.args.nbs \/ self.batch_size), 1)  # accumulate loss before optimizing\n        weight_decay = self.args.weight_decay * self.batch_size * self.accumulate \/ self.args.nbs  # scale weight_decay\n        iterations = math.ceil(len(self.train_loader.dataset) \/ max(self.batch_size, self.args.nbs)) * self.epochs\n        self.optimizer = self.build_optimizer(\n            model=self.model,\n            name=self.args.optimizer,\n            lr=self.args.lr0,\n            momentum=self.args.momentum,\n            decay=weight_decay,\n            iterations=iterations,\n        )\n        # Scheduler\n        # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565<\/code><\/pre>\n<p>\u6838\u5fc3\u4ee3\u78013\uff1a\u5f00\u59cb\u8bad\u7ec3<\/p>\n<pre><code class=\"language-python\">def _do_train(self, world_size=1):\n        &quot;&quot;&quot;Train completed, evaluate and plot if specified by arguments.&quot;&quot;&quot;\n        if world_size &gt; 1:\n            self._setup_ddp(world_size)\n        self._setup_train(world_size)\n\n        nb = len(self.train_loader)  # number of batches\n        nw = max(round(self.args.warmup_epochs * nb), 100) if self.args.warmup_epochs &gt; 0 else -1  # warmup iterations\n        last_opt_step = -1\n        self.epoch_time = None\n        self.epoch_time_start = time.time()\n        self.train_time_start = time.time()\n        # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n        epoch = self.start_epoch\n        self.optimizer.zero_grad()  # zero any resumed gradients to ensure stability on train start\n        while True:\n            self.epoch = epoch\n            self.run_callbacks(&quot;on_train_epoch_start&quot;)\n            # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n            self.model.train()\n            if RANK != -1:\n                self.train_loader.sampler.set_epoch(epoch)\n            pbar = enumerate(self.train_loader)\n            # Update dataloader attributes (optional)\n            if epoch == (self.epochs - self.args.close_mosaic):\n                self._close_dataloader_mosaic()\n                self.train_loader.reset()\n\n            # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n            self.tloss = None\n            for i, batch in pbar:\n                self.run_callbacks(&quot;on_train_batch_start&quot;)\n                # Warmup\n                # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n                # Forward \u6b63\u5411\u4f20\u64ad\n                with torch.cuda.amp.autocast(self.amp):\n                    batch = self.preprocess_batch(batch)\n                    self.loss, self.loss_items = self.model(batch)  # \u635f\u5931\u8ba1\u7b97\n                    if RANK != -1:\n                        self.loss *= world_size\n                    self.tloss = (\n                        (self.tloss * i + self.loss_items) \/ (i + 1) if self.tloss is not None else self.loss_items\n                    )\n\n                # Backward \u53cd\u5411\u4f20\u64ad\n                self.scaler.scale(self.loss).backward()\n\n                # Optimize \u4f18\u5316\u4e00\u6b65\n                if ni - last_opt_step &gt;= self.accumulate:\n                    self.optimizer_step()\n                    last_opt_step = ni\n\n                    # Timed stopping\n                    # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n                # Log\n                mem = f&quot;{torch.cuda.memory_reserved() \/ 1E9 if torch.cuda.is_available() else 0:.3g}G&quot;  # (GB)\n                loss_len = self.tloss.shape[0] if len(self.tloss.shape) else 1\n                losses = self.tloss if loss_len &gt; 1 else torch.unsqueeze(self.tloss, 0)\n                # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n            self.lr = {f&quot;lr\/pg{ir}&quot;: x[&quot;lr&quot;] for ir, x in enumerate(self.optimizer.param_groups)}  # for loggers\n            self.run_callbacks(&quot;on_train_epoch_end&quot;)\n            if RANK in {-1, 0}:\n                final_epoch = epoch + 1 &gt;= self.epochs\n                self.ema.update_attr(self.model, include=[&quot;yaml&quot;, &quot;nc&quot;, &quot;args&quot;, &quot;names&quot;, &quot;stride&quot;, &quot;class_weights&quot;])\n\n                # Validation\n                # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n                # Save model \u4fdd\u5b58\u6a21\u578b\n                if self.args.save or final_epoch:\n                    self.save_model()\n                    self.run_callbacks(&quot;on_model_save&quot;)\n\n            # Scheduler\n            # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565\n\n            # Early Stopping\n            if RANK != -1:  # if DDP training\n                broadcast_list = [self.stop if RANK == 0 else None]\n                dist.broadcast_object_list(broadcast_list, 0)  # broadcast &#039;stop&#039; to all ranks\n                self.stop = broadcast_list[0]\n            if self.stop:\n                break  # must break all DDP ranks\n            epoch += 1\n\n        # \u7bc7\u5e45\u539f\u56e0\uff0c\u4ee3\u7801\u5df2\u7701\u7565<\/code><\/pre>\n<p>\u4ee3\u7801\u89e3\u6790\uff1a<br \/>\n\u5728_do_train\u51fd\u6570\u4e2d\uff0c\u53ef\u4ee5\u770b\u5230\u6df1\u5ea6\u5b66\u4e60\u57fa\u672c\u7684\u6b65\u9aa4\uff0c\u5373\uff1a<\/p>\n<ol>\n<li>\u6b63\u5411\u4f20\u64ad<\/li>\n<li>\u635f\u5931\u8ba1\u7b97<\/li>\n<li>\u53cd\u5411\u4f20\u64ad<\/li>\n<li>\u4f18\u5316\u4e00\u6b65<\/li>\n<li>\u6e05\u7a7a\u68af\u5ea6<\/li>\n<li>\u4fdd\u5b58\u6a21\u578b<br \/>\n<blockquote>\n<p>\u5907\u6ce8\uff1a\u6e05\u7a7a\u68af\u5ea6\u5c01\u88c5\u5728optimizer_step\u51fd\u6570\u4e2d\u4e86\u3002<\/p>\n<\/blockquote>\n<\/li>\n<\/ol>\n<h2><span class=\"ez-toc-section\" id=\"%E5%86%85%E5%AE%B9%E5%B0%8F%E7%BB%93\"><\/span>\u5185\u5bb9\u5c0f\u7ed3<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<ul>\n<li>YOLOv8\u7684\u7f51\u7edc\u7ed3\u6784\n<ul>\n<li>\u4e3b\u8981\u6709\u4e3b\u5e72\u7f51\u7edc\u548cHead\u7f51\u7edc\u7ec4\u6210<\/li>\n<li>\u4e3b\u5e72\u7f51\u7edc\u4e2d\u8fdb\u884cP1-P5\u5c42\u5377\u79ef\uff0c\u7ecf\u8fc7SPPF\u540e\u8fdb\u884cUpsample\u4e0a\u91c7\u6837\u64cd\u4f5c\u540e\uff0c\u4e0eP3\u3001P4\u3001P5\u8fdb\u884cconcat\u64cd\u4f5c<\/li>\n<li>\u6700\u540e\u901a\u8fc715\u5c42\u300118\u5c42\u300121\u5c42\u5206\u522b\u7ecf\u8fc7Detect\u6a21\u5757\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\u3002<\/li>\n<\/ul>\n<\/li>\n<li>YOLOv8\u7684\u4ee3\u7801\u89e3\u6790\n<ul>\n<li>\u6784\u5efa\u6a21\u578b\u7684\u6838\u5fc3\u51fd\u6570\u662f\uff1aself._smart_load()\u3001self.task_map[self.task][key]\u548cparse_model\u51fd\u6570<\/li>\n<li>self.task_map\u4f7f\u7528\u4e86\u5b57\u7b26\u4e32\u952e\u4e0e\u7c7b\u5bf9\u8c61\u503c\u8fdb\u884c\u5173\u8054\uff0c\u7531\u6b64\u8fbe\u5230\u6839\u636e\u5173\u952e\u5b57\u9009\u62e9\u5bf9\u5e94\u7684\u6a21\u578b\u7c7b<\/li>\n<li>parse_model\u51fd\u6570\u4e2d\u901a\u8fc7\u8bfb\u53d6\u53c2\u6570\u3001\u8bbe\u7f6e\u6fc0\u6d3b\u51fd\u6570\u540e\uff0c\u4f7f\u7528nn.Sequential\u521b\u5efa\u5bf9\u5e94\u7684\u6a21\u578b<\/li>\n<li>\u8bad\u7ec3\u6a21\u578b\u7684\u6838\u5fc3\u51fd\u6570\u4e3aClassificationDataset\u7684\u5c01\u88c5\u3001_setup_train()\u51fd\u6570\u548c_do_train()\u51fd\u6570<\/li>\n<li>_do_train()\u51fd\u6570\u4e2d\u5305\u542b\u6df1\u5ea6\u5b66\u4e60\u7684\u57fa\u672c\u6b65\u9aa4\uff0c\u5373\uff1a\u6b63\u5411\u4f20\u64ad\u2192\u635f\u5931\u8ba1\u7b97\u2192\u53cd\u5411\u4f20\u64ad\u2192\u4f18\u5316\u4e00\u6b65\u2192\u6e05\u7a7a\u68af\u5ea6\u2192\u4fdd\u5b58\u6a21\u578b<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<h2><span class=\"ez-toc-section\" id=\"%E5%8F%82%E8%80%83%E8%B5%84%E6%96%99\"><\/span>\u53c2\u8003\u8d44\u6599<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><a href=\"https:\/\/www.bilibili.com\/video\/BV1tV411K7cp?vd_source=f825e70d1502bfc21582d29e60d89bb5\">B\u7ad9\uff1aYoloV8Ultralytics\u6a21\u578b\u7ed3\u6784\u8be6\u7ec6\u8bb2\u89e3<\/a><\/p>\n<p><a href=\"https:\/\/developer.aliyun.com\/article\/1430611\">YOLOv8\u6a21\u578byaml\u7ed3\u6784\u56fe\u7406\u89e3\uff08\u9010\u5c42\u5206\u6790\uff09<\/a><\/p>\n<p align=\"center\">\u6b22\u8fce\u5173\u6ce8\u516c\u4f17\u53f7\u4ee5\u83b7\u5f97\u6700\u65b0\u7684\u6587\u7ae0\u548c\u65b0\u95fb<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/09\/\u626b\u7801_\u641c\u7d22\u8054\u5408\u4f20\u64ad\u6837\u5f0f-\u767d\u8272\u7248.bmp\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/09\/\u626b\u7801_\u641c\u7d22\u8054\u5408\u4f20\u64ad\u6837\u5f0f-\u767d\u8272\u7248.bmp\" alt=\"\" \/><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u524d\u8a00 \u5728\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day11\uff08\u4e0b\uff09\uff1aY [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":2339,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"aside","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","theme-transparent-header-meta":"default","adv-header-id-meta":"","stick-header-meta":"default","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[28],"tags":[65,66],"class_list":["post-2367","post","type-post","status-publish","format-aside","has-post-thumbnail","hentry","category-blog","tag-65","tag-66","post_format-post-format-aside"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.4 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day12\uff1aYOLO\u7684\u6df1\u5165\u4e86\u89e3 - \u4e00\u8d77AI\u6280\u672f<\/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:\/\/17aitech.com\/?p=2367\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/17aitech.com\/?p=2367\",\"url\":\"https:\/\/17aitech.com\/?p=2367\",\"name\":\"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day12\uff1aYOLO\u7684\u6df1\u5165\u4e86\u89e3 - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\/\/17aitech.com\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/17aitech.com\/?p=2367#primaryimage\"},\"image\":{\"@id\":\"https:\/\/17aitech.com\/?p=2367#primaryimage\"},\"thumbnailUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLO\u5c01\u9762.jpg\",\"datePublished\":\"2024-06-26T08:17:41+00:00\",\"dateModified\":\"2024-10-08T07:18:03+00:00\",\"author\":{\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739\"},\"breadcrumb\":{\"@id\":\"https:\/\/17aitech.com\/?p=2367#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/17aitech.com\/?p=2367\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/?p=2367#primaryimage\",\"url\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLO\u5c01\u9762.jpg\",\"contentUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLO\u5c01\u9762.jpg\",\"width\":480,\"height\":360},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/17aitech.com\/?p=2367#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/17aitech.com\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day12\uff1aYOLO\u7684\u6df1\u5165\u4e86\u89e3\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/17aitech.com\/#website\",\"url\":\"https:\/\/17aitech.com\/\",\"name\":\"\u4e00\u8d77AI\u6280\u672f\",\"description\":\"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca\",\"alternateName\":\"\u4e00\u8d77AI\u6280\u672f\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/17aitech.com\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739\",\"name\":\"Dongming\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/image\/\",\"url\":\"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg\",\"contentUrl\":\"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg\",\"caption\":\"Dongming\"},\"description\":\"\u89c1\u5929\u5730\uff0c\u89c1\u4f17\u751f\uff0c\u89c1\u81ea\u5df1\u3002\",\"sameAs\":[\"http:\/\/17aitech.com\"],\"url\":\"https:\/\/17aitech.com\/?page_id=33738&user=1\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day12\uff1aYOLO\u7684\u6df1\u5165\u4e86\u89e3 - \u4e00\u8d77AI\u6280\u672f","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:\/\/17aitech.com\/?p=2367","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=2367","url":"https:\/\/17aitech.com\/?p=2367","name":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day12\uff1aYOLO\u7684\u6df1\u5165\u4e86\u89e3 - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=2367#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=2367#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLO\u5c01\u9762.jpg","datePublished":"2024-06-26T08:17:41+00:00","dateModified":"2024-10-08T07:18:03+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=2367#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=2367"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=2367#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLO\u5c01\u9762.jpg","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/06\/YOLO\u5c01\u9762.jpg","width":480,"height":360},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=2367#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"\u3010\u8bfe\u7a0b\u603b\u7ed3\u3011Day12\uff1aYOLO\u7684\u6df1\u5165\u4e86\u89e3"}]},{"@type":"WebSite","@id":"https:\/\/17aitech.com\/#website","url":"https:\/\/17aitech.com\/","name":"\u4e00\u8d77AI\u6280\u672f","description":"\u8ba9AI\u77e5\u8bc6\u89e6\u624b\u53ef\u53ca","alternateName":"\u4e00\u8d77AI\u6280\u672f","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/17aitech.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/17aitech.com\/#\/schema\/person\/3d23bb6f7f115fcefc9ae7803a691739","name":"Dongming","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/#\/schema\/person\/image\/","url":"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg","contentUrl":"\/\/17aitech.com\/wp-content\/uploads\/member\/avatars\/238a0b923820dcc5.1732798681.jpg","caption":"Dongming"},"description":"\u89c1\u5929\u5730\uff0c\u89c1\u4f17\u751f\uff0c\u89c1\u81ea\u5df1\u3002","sameAs":["http:\/\/17aitech.com"],"url":"https:\/\/17aitech.com\/?page_id=33738&user=1"}]}},"_links":{"self":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/2367","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2367"}],"version-history":[{"count":8,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/2367\/revisions"}],"predecessor-version":[{"id":33241,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/2367\/revisions\/33241"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/media\/2339"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2367"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2367"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2367"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}