{"id":273,"date":"2024-10-23T17:25:09","date_gmt":"2024-10-23T09:25:09","guid":{"rendered":"https:\/\/eve2333.top\/?p=273"},"modified":"2024-10-23T17:25:10","modified_gmt":"2024-10-23T09:25:10","slug":"yolov5%e5%ae%9e%e6%88%98%e6%8b%93%e5%b1%95","status":"publish","type":"post","link":"https:\/\/eve2333.top\/?p=273","title":{"rendered":"YOLOv5\u5b9e\u6218\u62d3\u5c55"},"content":{"rendered":"\n

AutoDL\u670d\u52a1\u5668\u90e8\u7f72\u8bad\u7ec3<\/h1>\n\n\n\n

\u6ce8\u518c\u8fc7\u7a0b\u4e0d\u4e88\u663e\u793a\uff0c\u7b80\u5355\u7684yolov5\u5176\u5b9e\u4e0d\u9700\u8981\u4ec0\u4e48\u9ad8\u7ea7\u5361\uff0c\u6bd5\u7adf\u8fdc\u90531050-4G\u7248\uff0c\u8fd1\u52304060-8g\u90fd\u53ef\u4ee5\u90e8\u7f72<\/p>\n\n\n\n

\u9009\u62e9\u540e\uff0c\u770b\u5230\u955c\u50cf\"\"<\/p>\n\n\n\n

\u4e0d\u8981\u9009\u62e9\uff01\uff01\uff01\u56e0\u4e3a\u8bad\u7ec3\u7684\u5206\u6570\u662f0 \uff0c\u9009\u62e9\u57fa\u7840\u7248\u672c\uff0cpytorch1.9 ubantu\u6269\u5927\u7248\u672c11.1\u4ee5\u4e0a<\/p>\n\n\n\n

\u7acb\u523b\u521b\u5efa\uff01\uff01\uff01jupyter lab\u7684\u65b9\u5f0f\u6253\u5f00\uff0c\u76f4\u63a5\u628agithub\u7684yolov5\u7684\u6587\u4ef6\u5939\u62d6\u5165\u5373\u53ef<\/p>\n\n\n\n

\u5982\u679c\u662fzip\uff0c\u7ec8\u7aef\u8f93\u5165 unzip \u6587\u4ef6\u5939\u540d \u89e3\u538b                    cd\u8fdb\u5165\u540e\uff0c\u548c\u539f\u6765\u4e00\u6837\u73af\u5883\u7b49\u7b49\u7b49\u7b49<\/p>\n\n\n\n

conda activate base\nconda init<\/code><\/pre>\n\n\n\n
\u9000\u51fa\u540e\u518d\u8fdb\u5165\u5373\u53ef<\/p>\n\n\n\n
pip install -r requirements.txt<\/code><\/pre>\n\n\n\n
\u6839\u636e\u4fe1\u606f\u5408\u7406\u4fee\u6539txt\u5185\u5bb9\uff0c\u53ef\u4ee5\u5c1d\u8bd5\u5c06\u5904\u7406\u7684pt\u6587\u4ef6\u4e0a\u4f20\u6d4b\u8bd5\u662f\u5426\u6b63\u5e38<\/p>\n\n\n\n
python detect.py<\/code><\/pre>\n\n\n\n
\u540c\u4e0a\u4e00\u7ae0\u64cd\u4f5c\u65b9\u6cd5<\/p>\n\n\n\n
Downloading https:\/ultralytics.com\/assets\/Arial.ttf to\/root\/.config\/Ultralytics\/Arial.ttf...<\/p>\n\n\n\n
\u53ef\u4ee5\u9009\u62e9\u76f4\u63a5\u4e0a\u4f20\u5b57\u4f53\u5230\u6587\u4ef6\u5939\u4e2d\uff0c\u8282\u7701\u62a5\u9519<\/p>\n\n\n\n
mv Arial.ttf \/root\/.config\/U1tralytics\/\npython train.py\ntar -cvf runs.tar.gz runs\/<\/code><\/pre>\n\n\n\n
\u5c31\u53ef\u4ee5\u4e0b\u8f7d\u4e86<\/p>\n\n\n\n
pycharm\u57fa\u672c\u4f7f\u7528\u65b9\u6cd5<\/h1>\n\n\n\n
\u81ea\u5df1\u627e\u4e00\u4e2apycharm professional\u7834\u89e3\u7248\u6216\u8005\u4e3a\u4ed6\u82b1\u70b9\u94b1<\/p>\n\n\n\n
\u672c\u4eba\u4e0b\u8f7d\u7684\u662f2023.3\u7248\u672c\uff0c\u521b\u5efa\u7f16\u8bd1\u5668\u73af\u5883\uff0c\u9009\u62e9\u4f60\u7684conda\u5373\u53ef<\/p>\n\n\n\n
\u5c31\u662f\u53ef\u4ee5\u66f4\u76f4\u63a5\u7684\u627e\u5230\u90a3\u4e2aconda\u7684python.exe<\/p>\n\n\n\n
\u6709\u4e00\u4e2aNLP\u5e38\u7528\u7684\u5305<\/p>\n\n\n\n
pip install jieba<\/code><\/pre>\n\n\n\n
\u5728\u6587\u4ef6\u4e2dimport jieba\uff0c\u53d1\u73b0\u8fd4\u56de\u4e86'\u6ca1\u6709\u627e\u5230jieba\u5305'<\/p>\n\n\n\n
\u8fd9\u662f\u56e0\u4e3a\u9ed8\u8ba4\u7684terminal\u5728windows\u7684powershell\u8fd0\u884c\uff0c\u548ccmd\u6709\u70b9\u95ee\u9898\uff0c\u5c06\u56fe\u7eb8\u5185\u5bb9\u6539\u4e3acmd\u5373\u53ef \u200b<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n
 \u5982\u4f55\u94fe\u63a5auto\u670d\u52a1\u5668\uff1f<\/h3>\n\n\n\n
\u63d0\u524d\u5f00\u673aauto\uff0c\u6709\u4e00\u4e2aSSH\u767b\u5f55\u6307\u4ee4\u5982\u4e0b\u586b\u5199\u3002      test\u586b\u5199\u5373\u53ef<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\u200b<\/p>\n\n\n\n
 \u4fee\u6539\u76f4\u63a5\u5728\u672c\u5730\u4fee\u6539\u4ee3\u7801\uff0c\u81ea\u52a8\u4e0a\u4f20\u5728tools----deployment-----Automatic---Update<\/p>\n\n\n\n
\u81ea\u52a8\u83b7\u53d6tools----deployment-----brows remote host\uff0c\u5c31\u53ef\u4ee5\u770b\u5230\u670d\u52a1\u5668\u91cc\u9762\u6587\u4ef6\uff0cdownload forhere\u5373\u53ef<\/p>\n\n\n\n
\u60f3\u8dd1\u4e00\u4e9b\u53c2\u6570,\u7f16\u8f91\u5c31\u8fde\u5230\u7ec8\u7aef\u4e0a\u9762
\"\"\"\"<\/p>\n\n\n\n
\u5982\u679c \u60f3\u8c03\u6574\u670d\u52a1\u5668\u548c\u672c\u5730\u7f16\u8bd1\u5668\uff0c\u53f3\u4e0a\u89d2\"\"\u70b9\u51fbEdit Configurations \u76f4\u63a5\u66f4\u6362\u5373\u53ef \u200b<\/p>\n\n\n\n
\u200bVsCode\u57fa\u672c\u4f7f\u7528\u65b9\u6cd5<\/h1>\n\n\n\n
\u76f4\u63a5\u5b89\u88c5\u540e\uff0c\u9009\u62e9python\u7684\u63d2\u4ef6\uff0c\u9009\u62e9\u89e3\u91ca\u5668\uff0c\u5373\u53f3\u4e0b\u89d2\uff0c\u9009\u62e9\u4f60\u7684conda\u73af\u5883\u5c31OK<\/p>\n\n\n\n
\u7531\u4e8epowershell\u7684\u5f71\u54cd\uff0c\u5728\u53f3\u8fb9\u6709\u4e00\u4e2a\u9009\u9879\u9009\u62e9\u9ed8\u8ba4\u914d\u7f6e\u6587\u4ef6\u6362\u6210cmd<\/p>\n\n\n\n
\u5982\u4f55\u5efa\u7acb\u4e00\u4e2a\u670d\u52a1\u5668\u5462\uff1f<\/p>\n\n\n\n
\u5b89\u88c5remote-ssh\u63d2\u4ef6\uff0cconnect to host\uff0cadd new ssh host\uff0ccopy\u4e00\u4e0bauto\u670d\u52a1\u5668\u4e0a\u7684ssh\uff0c\u5bc6\u7801\u5373\u53ef<\/p>\n\n\n\n
vscode\u5176\u5b9e\u662f\u76f4\u63a5\u5904\u7406\u670d\u52a1\u5668\uff0c\u56e0\u6b64\u5c31\u50cf\u7f51\u9875\u7aef\u7684Auto\u4f7f\u7528\u5373\u53ef,\u7b2c\u4e8c\u6b21\u6253\u5f00\u5f88\u5bb9\u6613\u51fa\u6765<\/p>\n\n\n\n
YOLOV5\u6a21\u578b\u7ed3\u6784\u548c\u6784\u5efa\u539f\u7406<\/h1>\n\n\n\n
\u7ed3\u6784\u5b9a\u4e49\u5728\u6587\u4ef6\u5939\u4e2d<\/p>\n\n\n\n
1.models\/yolov5*.yaml              2.models\/yolo.py           3.models\/commons.py  <\/p>\n\n\n\n
\u6bd4\u5982\u6211\u4eec\u6253\u5f00yolov5s.yaml\u7684\u6587\u4ef6\uff0c\u5982\u6b64\u4ee3\u7801\u501f\u52a9\u6ce8\u91ca\u548cai\u53ef\u4ee5\u77e5\u9053\u5230\u5e95\u662f\u5e72\u4ec0\u4e48\u7528\u7684<\/p>\n\n\n\n
\u5b9e\u5728\u4e0d\u7406\u89e3\u63d0\u4f9bTensorboard\u53ef\u89c6\u5316\u5e2e\u52a9\uff0c\uff0c\u5728\u7ec8\u7aef\u8f93\u5165\u5982\u4e0b\u4ee3\u7801<\/p>\n\n\n\n
tensorboard --logdir runs\u00a0<\/code><\/pre>\n\n\n\n
\u70b9\u51fb GRAPHS \u5c31\u53ef\u4ee5\u770b\u89c1\u4e86\uff0c\u90a3\u4ed6\u73b0\u5728\u7684\u8bdd\u662f\u975e\u5e38\u7b80\u5355\u7684\uff0c\u5c31\u662f\u53ea\u6709\u4e00\u4e2a\u8f93\u5165\uff0c\u7136\u540e\u4e00\u4e2a detection model\uff0c\u7136\u540e\u4e00\u4e2a output\uff0c\u7136\u540e\u6211\u4eec\u53ea\u8981\u53cc\u51fb\u8fd9\u4e2a detection model\uff0c\u5b83\u5c31\u53ef\u4ee5\u628a\u6574\u4e2a\u7f51\u7edc\u7ed3\u6784\u5c55\u5f00\u4e86\uff0c\u7136\u540e\u6211\u7684\u9f20\u6807\u6eda\u8f6e\u7684\u8bdd\uff0c\u62d6\u6eda\u8f6e\u6eda\u52a8\u7684\u8bdd\uff0c\u5b83\u5c31\u662f\u653e\u5927\u7f29\u5c0f\uff0c\u7136\u540e\u6309\u4f4f\u7684\u8bdd\u62d6\u52a8\u5b83\uff0c\u5c31\u4f1a\u628a\u8fd9\u4e2a\u56fe\u5c31\u53ef\u4ee5\u8fdb\u884c\u62d6\u52a8\u4e86\uff0c\u90a3\u6211\u4eec\u73b0\u73b0\u5728\u7684\u8bdd\u5c31\u5bf9\u7740\u8fd9\u4e2a\u56fe\u4ee5\u53ca\u521a\u624d\u7684\u8fd9\u4e2a\u7ed3\u6784\uff0c\u7136\u540e\u6765\u628a YOLOV \u6211\u7684\u6574\u4f53\u7684\u7ed3\u6784\u8fc7\u4e00\u4e0b\uff0c\u8fc7\u6e21\u8fc7\u7a0b\u4e2d\u5462\uff0c\u6211\u4e5f\u4f1a\u6839\u636e tt bot \u4e2d\u7684\u8fd9\u4e2a\u63d0\u793a\u554a\u6216\u8005\u8fd9\u4e2a\u4fe1\u606f\u5427\uff0c\u544a\u8bc9\u5927\u5bb6\uff0c\u5c31\u662f\u8fd9\u4e2a\u6570\u636e\u5728\u7f51\u7edc\u4e2d\u7ecf\u8fc7\u6bcf\u4e00\u5c42\u4e4b\u540e\uff0c\u5b83\u90fd\u662f\u600e\u4e48\u53d8\u6362\u7684\uff0cOK \u90a3\u8ba9\u6211\u4eec\u5f00\u59cb\uff0c\u90a3\u9996\u5148\u524d\u9762\u53ef\u80fd\u6bd4\u8f83\u7b80\u5355\u4e86\uff0c\u5728 BACKBALL \u91cc\u9762\u7684\u8bdd\uff0c\u4f60\u4f1a\u53d1\u73b0\u4ed6\u524d\u9762\u7684\u8bdd\u5c31\u662f\u6309\u7167\u987a\u5e8f\u8d70\u4e0b\u6765\uff0c\u6bcf\u4e00\u5c42\u7684\u8fd9\u4e2a\u8f93\u5165\u90fd\u662f\u4e00\u5bf9\u5427\uff0c\u90a3\u5176\u5b9e\u8bf4\u767d\u4e86\u5c31\u662f\u4e0a\u4e00\u5c42\u7684\u8f93\u51fa\u5c31\u662f\u5f53\u524d\u5c42\u7684\u8f93\u5165\uff0c\u90a3\u6211\u4eec\u770b\u4e00\u4e0b\u5b9e\u9645\u7684\u8fd9\u4e2a\u7ed3\u6784\uff0c\u90a3\u6211\u4eec\u6709\u4e00\u4e2a\u8f93\u5165 input \u5c31\u662f\u539f\u59cb\u7684\u8fd9\u4e2a\u56fe\u7247\uff0c\u90a3\u5b83\u7684\u7ed3\u6784\u5b83\u7684\u7ef4\u5ea6\u7684\u8bdd\u5c31\u662f 1\u00d73 \u4e58 640\u00d7640\uff0c\u7136\u540e\u7d27\u63a5\u7740\u8f93\u5165\u7b2c\u4e00\u5c42 CNV0\uff0c\u90a3\u8fd9\u91cc\u7684\u8bdd\u53ef\u4ee5\u770b\u5230\u5b83\u7684\u8f93\u5165\u662f 1\u00d73 \u4e58\u4ee5 640\u00d7640\uff0c\u7136\u540e\u7ecf\u8fc7\u5b83\u4e4b\u540e\uff0c\u6574\u4e2a\u7684\u8fd9\u4e2a\u7279\u5f81\u5c31\u53d8\u6210\u4e86 1\u00d732 \u4e58\u4e09\u4e8c 0\u00d7320\uff0c\u53ef\u4ee5\u770b\u5230\u5b83\u6bd4\u4ee5\u524d\u7f29\u5c0f\u4e86 1\/2 \u5bf9\u5427\uff0c\u90a3\u8fd9\u4e2a\u600e\u4e48\u5728\u4ee5\u524d\u5728\u8fd9\u4e2a\u538b\u7f29\u6587\u4ef6\u91cc\u9762\u4f53\u73b0\u5462\uff0c\u6240\u4ee5\u5c31\u5728\u8fd9\u91cc\u9762\uff0c\u5c31\u4ed6\u6700\u540e\u9664\u4ee5\u4e8c\uff0c\u5c31\u610f\u601d\u5c31\u662f\u544a\u8bc9\u6211\u4eec\uff0c\u6211\u4eec\u73b0\u5728\u6574\u4e2a\u7279\u5f81\u56fe\u53d8\u4e3a\u4e86\u539f\u6765\u7684 1\/2\uff0c\u7136\u540e\u6211\u4eec\u518d\u5f80\u4e0b\u770b\uff0c\u5176\u5b9e\u5b83\u5728\u91cc\u9762\u6807\u8bb0\u4e86\u5f88\u591a\u5904\u554a\uff0c\u5c31\u9664\u4ee5 4\u00f78\uff0c\u9664\u4ee5 16\uff0c\u9664\u4ee5 32 \u5bf9\u5427\uff0c\u6211\u4eec\u53ef\u4ee5\u770b\u4e00\u4e0b\uff0c\u90a3\u4e0b\u4e00\u6b21\u4ed6\u8bf4\u4e0b\u6b21\u53c8\u53d8\u6210\u9664\u4ee5\u56db\uff0c\u90a3\u8bf4\u660e\u7279\u5f81\u56fe\u53c8\u53d8\u5c0f\u4e86\uff0c\u90a3\u4e48\u539f\u59cb\u503c 640\u00d7640\uff0c\u7136\u540e\u7ecf\u8fc7\u4e00\u6b21\u9664\u4ee5\u4e8c\uff0c\u53d8\u6210 320\u00d7320\uff0c\u7136\u540e\u518d\u9664\u4ee5\u56db\uff0c\u7136\u540e\u6211\u4eec\u518d\u6765\u770b\u4e00\u4e0b\uff0c\u5c31\u8fd9\u4e2a CV \u4e00\u70b9\u4e00\u4e0b\u53ef\u4ee5\u770b\u5230\uff0c\u90a3\u8fd9\u4e2a\u65f6\u5019\u4ed6\u7684\u8fd9\u4e2a\u8f93\u5165\u5c31\u662f 1\u00d732 \u4e58 320\u00d7320\uff0c\u4ed6\u7684\u8fd9\u4e2a\u8f93\u51fa\u7684\u8bdd\u5c31\u662f 1\u00d764 \u4e58 100\uff0c6\u00d7160 \u5bf9\u5427\uff0c\u53ef\u4ee5\u770b\u5230\u5b83\u786e\u5b9e\u662f\u53d8\u6210\u4e86\u5443\u539f\u59cb\u56fe\u50cf\u7684\u8fd9\u4e2a\u591a\u5c11\u5206\u4e4b\u4e00\uff0c\u8fd9\u91cc\u7684\u8bdd\u8fd8\u8981\u518d\u6570\u70b9\uff0c\u5c31\u770b\u5b83\u5b83\u524d\u9762\u662f\u4ec0\u4e48\u610f\u601d\u5462\uff0c\u524d\u9762\u8fd9\u4e2a\u96f6\u7684\u8bdd\uff0c\u5927\u5bb6\u53ef\u4ee5\u7406\u89e3\u5c31\u662f\u4ed6\u662f\u5bf9\u8fd9\u4e2a\u7f51\u7edc\u7684\u4e00\u4e2a\u7f16\u53f7\uff0c\u4e3a\u4ec0\u4e48\u4ed6\u6709\u8fd9\u4e48\u4e00\u4e2a\u7f16\u53f7\u5462\uff0c\u662f\u56e0\u4e3a\u4ed6\u540e\u9762\u7684\u8bdd\u8981\u7528\u5230\uff0c\u56e0\u4e3a\u5b83\u7684\u867d\u7136\u524d\u9762 backbone \u7684\u8fd9\u4e2a\u7f51\u7edc\u7ed3\u6784\uff0c\u662f\u4e0a\u4e00\u5c42\u7684\u8f93\u51fa\u5c31\u662f\u4e0b\u4e00\u5c42\u7684\u8f93\u5165\uff0c\u4f46\u662f\u4ed6\u5728\u540e\u9762\u7684\u65f6\u5019\uff0c\u5b83\u5c31\u4e0d\u662f\u8fd9\u6837\u7684\u7ed3\u6784\u4e86\uff0c\u5b83\u53ef\u80fd\u4f1a\u5f15\u7528\u524d\u9762\u7684\u90a3\u4e2a\uff0c\u4ed6\u53ef\u80fd\u4f1a\u628a\u524d\u51e0\u5c42\u7684\u8f93\u51fa\u8fde\u63a5\u76f4\u63a5\u8fde\u5230\u540e\u9762\u53bb\uff0c\u5c31\u662f\u4e0d\u7ecf\u8fc7\u5176\u4ed6\u7684\u64cd\u4f5c\uff0c\u76f4\u63a5\u8fde\u5230\u540e\u9762\u53bb\uff0c\u6240\u4ee5\u8bf4\u4ed6\u8981\u5bf9\u4ed6\u8fdb\u884c\u8bb0\u5f55\uff0c\u5305\u62ec\u540e\u9762\u6211\u4eec\u5728\u6539\u7ed3\u6784\u7684\u65f6\u5019\uff0c\u8fd9\u4e2a\u7f16\u53f7\u4e5f\u662f\u975e\u5e38\u91cd\u8981\u7684\uff0c\u6211\u4eec\u8981\u628a\u5b83\u8bb0\u6e05\u695a\uff0c\u90a3\u7b2c\u4e00\u5c42\u7684\u8bdd\u5c31\u662f\u4ece\u96f6\u5f00\u59cb\u6807\uff0c\u7136\u540e 01234567 \u8fd9\u6837\u8d70\u4e0b\u6765\uff0c\u7136\u540e\u5176\u5b9e\u53ef\u4ee5\uff0c\u6211\u4eec\u5728\u8fd9\u4e2a\u56fe\u4e0a\u4e5f\u662f\u53ef\u4ee5\u6e05\u6670\u770b\u5230\u7684\uff0c\u90a3\u6211\u7684\u8fd9\u4e2a\u6bd4\u5982\u8fd9\u4e2a\u7b2c\u4e00\u5c42\u5377\u79ef\uff0c\u90a3\u5c31\u662f\u96f6\uff0c\u7b2c\u4e8c\u5c42\u5377\u79ef\u5c31\u662f\u4e00\uff0c\u7b2c\u4e09\u5c42\u5c31\u662f\u7b2c\u4e09\u5957\uff0c\u8fd9\u4e2a C3 \u5c31\u662f\u4e8c\u5bf9\u5427\uff0c\u5c31\u4ed6\u8fd9\u91cc\u6ca1\u6709\u6807\u51fa\u6765\u7684\u8fd9\u4e2a\u4e8c\uff0c\u90a3\u7d27\u63a5\u7740\uff0c\u7136\u540e\u6211\u4eec\u7ee7\u7eed\u518d\u5f80\u4e0b\u8d70\uff0c\u53c8\u662f\u4e00\u5c42\u7684\u8fd9\u4e2a\u5377\u79ef\u5bf9\u5427\uff0c\u90a3\u53ef\u4ee5\u770b\u5230\u90a3\u8fd9\u4e00\u5c42\u7684\u65f6\u5019\uff0c\u4ed6\u7684\u8f93\u51fa\u5c31\u662f 160\u00d7160\uff0c\u7136\u540e\u4ed6\u5728\u8f93\u51fa\u7684\u65f6\u5019\u554a\uff0c\u8fd9\u91cc\u53ef\u80fd\u5b83\u56e0\u4e3a\u5b83\u8f93\u51fa\u4e86\u4e24\u4e2a tensor\uff0c\u7136\u540e\u4ed6\u4fe9\u5c31\u6ca1\u6807\u51fa\u6765\uff0c\u6211\u4eec\u53ef\u4ee5\u628a\u5b83\u55ef\u53d6\u6d88\u4e00\u4e0b\uff0c\u53cc\u51fb\u7684\u8bdd\u5c31\u53ef\u4ee5\u6253\u5f00\u8fd9\u4e2a\u7f51\u7edc\u7684\u7ec6\u8282\u554a\uff0c\u6211\u4eec\u53ef\u4ee5\u53cc\u51fb\u5b83\uff0c\u7136\u540e\u53ef\u4ee5\u770b\u5230\u4ed6\u8fd9\u4e00\u6b65\u6709\u8d8a\u6765\u8d8a\u6a21\u7cca\uff0c\u4f46\u662f\u53ef\u4ee5\u770b\u5230\u5b83\u6700\u7ec8\u53d8\u6210\u4e86\uff0c\u5c31\u662f 128\u00d780 \u4e58 80 \u8fd9\u6837\u7684\u4e00\u4e2a\u7ed3\u6784\uff0c\u73b0\u5728\u5c31\u662f\u53c8\u5f80\u4e0b\u7f29\u4e86\u4e00\u5c42\uff0c\u6b63\u89c4\u56db\u5377\u79ef\u7684\u8bdd\uff0c\u4ed6\u5c31\u5f80\u4e0b\uff0c\u4ed6\u5c31\u505a\u4e86\u4e00\u6b21\uff0c\u5f53\u4e8e\u505a\u4e86\u4e00\u6b21\u4e0b\u91c7\u6837\uff0c\u7136\u540e\u518d\u7ee7\u7eed\u5f80\u4e0b\u8d70\u5230 C3 \u7684\u8fd9\u4e2a\u7b2c\u56db\u5c42 C3\uff0c\u7136\u540e\u4e00\u6b21\u4e00\u6b21\u4f60\u4f1a\u53d1\u73b0\uff0c\u7136\u540e\u5230\u7b2c\u56db\u5c42\u7684\u65f6\u5019\uff0c\u5b83\u5c31\u5206\u53c9\u4e86\uff0c\u90a3\u6211\u4eec\u5148\u4e0d\u8d70\u5b83\u5206\u53c9\u8fd9\u4e2a\u8def\u7ebf\uff0c\u6211\u4eec\u5148\u6309\u7167 backbone \u8fd9\u4e2a\u8def\u7ebf\u628a\u5b83\u8d70\u4e0b\u53bb\uff0c\u5c31\u662f C3\uff0c\u7d27\u63a5\u7740\u540e\u9762\u518d\u63a5\u4e00\u5c42\u5377\u79ef\uff0c\u7136\u540e\u5230\u8fd9\u4e2a\u7b2c\u4e94\u5c42\uff0c\u7136\u540e\u518d C3 \u5230\u7b2c\u516d\u5c42\uff0c\u90a3\u8fd9\u6bb5\u8bdd\uff0c\u7136\u540e\u518d\u7ee7\u7eed\uff0c\u7136\u540e\u5230\u7b2c\u4e03\u5c42\uff0c\u7b2c\u516b\u5c42\uff0c\u7136\u540e\u63a5\u4e0b\u6765\u7d27\u63a5\u7740\u8fc7\u4e86\u4e00\u4e2a SPPF \u662f\u7b2c\u4e5d\u5c42\uff0c\u76f4\u5230\u8fd9\u4e2a\u7b2c\u4e5d\u5c42\u554a\uff0c\u6216\u8005\u8bf4\u5e94\u8be5\u662f\u6574\u4e2a\u7684\u7b2c\u5341\u5c42\u4e86\u554a\uff0c\u5c31\u662f\u6b63\u597d\u5e94\u8be5\u662f\u6211\u4eec\u8fd9\u4e2a backbone \u7ed3\u675f\u7684\u8fd9\u4e00\u5c42\uff0c\u90a3\u6211\u4eec\u53ef\u4ee5\u770b\u554a\uff0c\u5c31\u5b83\u6709\u51e0\u4e2a\u5173\u952e\u7684\u70b9\uff0c\u4ed6\u505a\u4e86\u4e00\u4e9b\u7279\u5f81\u7684\u53d8\u6362\uff0c\u5c31\u662f\u6bcf\u6b21\u5377\u79ef\u7684\u8bdd\uff0c\u5c31\u6bcf\u4e00\u6b21\u5355\u72ec\u7684\u5377\u79ef\u5c42\uff0c\u5176\u5b9e\u4ed6\u90fd\u662f\u505a\u4e86\u4e00\u6b21\u4e0b\u91c7\u6837\uff0c\u4ece\u7b2c\u4e00\u6b21\u5377\u79ef\uff0c\u7136\u540e\u8ba9\u5b83\u53d8\u6210\u6574\u4f53\u7684\u4e00\u534a\uff0c\u5c31\u662f\u5c31\u662f 640 \u518d\u9664\u4ee5\u4e8c\u4e09\u767e\u4e8c\uff0c\u7136\u540e\u518d\u4e00\u6b21\u5377\u79ef\uff0c\u7136\u540e\u518d\u9664\u4ee5\u4e8c\u53d8\u6210 160\uff0c\u7136\u540e\u518d\u4e00\u6b21\u662f 80\uff0c\u518d\u8fc7\u4e00\u6b21\u662f 40\uff0c\u7136\u540e\u7ecf\u8fc7 CNV7 \u4e4b\u540e\uff0c\u5b83\u73b0\u5728\u5c31\u53d8\u6210\u4e86 20\u00d720 \u4e86\uff0c\u5bf9\u8fd8\u6709\u4e2a\u7279\u5f81\u56fe\u662f\u8fd9\u6837\u53d8\u6362\u7684\uff0cOK \u90a3\u6211\u4eec\u7ee7\u7eed\u5f80\u4e0b\uff0c\u7ee7\u7eed\u5f80\u4e0b\u8d70\u7684\u8bdd\uff0c\u5c31\u5230\u4e86\u8fd9\u4e2a head \u7684\u90e8\u5206\uff0chead \u7684\u90e8\u5206\u7684\u8bdd\u5443\uff0c\u5176\u5b9e\u89e3\u6790\u4e0b\u80af\u5b9a\u524d\u9762\u662f\u7c7b\u4f3c\u7684\uff0c\u5c31\u662f\u8bf4\u4ed6\u7b2c\u4e00\u5c42\u8fd9\u4e2a head\uff0c\u6211\u4eec\u4e5f\u662f\u6309\u7167\u8fd9\u4e2a\u5e8f\u53f7\u8d70\u5427\uff0c\u4ed6\u7b2c\u4e00\u5c42\u7684\u8fd9\u4e2a head \u5176\u5b9e\u5c31\u662f\u5148\u8fd8\u662f\u4e00\u4e2a\u5377\u79ef\uff0c\u5c31\u5bf9\u5e94\u7684\u662f\u56fe\u91cc\u9762\u7684\u8fd9\u4e2a\u4f4d\u7f6e\uff0c\u7136\u540e\u7d27\u63a5\u7740\u4e00\u6b21 on sample\uff0c\u5c31\u662f\u4e00\u6b21\u4e0a\u91c7\u6837 on sample\uff0c\u7136\u540e\u4e0b\u9762\u7684\u8bdd\u4ed6\u505a\u4e86\u4e00\u6b21\u62fc\u63a5\uff0c\u53ef\u4ee5\u770b\u5230\u5b83\u8fd9\u91cc\u7684\u8bdd\u4ed6\u544a\u8bc9\u4f60\uff0c\u4ed6\u8981\u628a\u8fd9\u4e2a backbone p4 \u7684\u8fd9\u4e2a\u5730\u65b9\u7ed9\u4ed6\u62fc\u8fc7\u6765\uff0c\u90a3\u6240\u4ee5\u8bf4\u5b83\u91cc\u7684\u8f93\u5165\u7684\u8bdd\u5c31\u53d8\u6210\u4e86\u4e00\u4e2a list\uff0c\u90a3\u5c31\u662f\u8bf4\u8fd9\u91cc\u7684\u662f\u4e00\uff0c\u662f\u6307\u4e0a\u4e00\u5c42\u516d\u7684\u8bdd\uff0c\u5c31\u662f\u8bf4\u6211\u7b2c\u516d\u5c42\u7684\u8fd9\u4e2a\u8f93\u51fa\uff0c\u7136\u540e\u505a\u4e86\u4e00\u4e2a\u62fc\u63a5\uff0c\u90a3\u6211\u4eec\u6765\u770b\u4e00\u4e0b\u662f\u4e0d\u662f\u539f\u59cb\u4e5f\u662f\u8fd9\u6837\uff0c\u5c31\u8bf4\u4ed6\u5728\u8fd9\u4e00\u6b21\u7684\u65f6\u5019\uff0c\u53ef\u4ee5\u770b\u5230\uff0c\u5b83\u662f\u628a\u7b2c\u516d\u5c42 C3 \u7684\u8fd9\u4e2a\u8f93\u51fa\u505a\u4e86\u4e00\u4e0b\u62fc\u63a5\uff0c\u90a3\u4e3a\u4ec0\u4e48\u4ed6\u4eec\u53ef\u4ee5\u62fc\u5728\u4e00\u8d77\u5462\uff0c\u662f\u56e0\u4e3a\u5443 C3 \u4e4b\u540e\u7684\u5c31\u662f\u8fd9\u4e2a CNV7\uff0c\u5c31\u662f\u7b2c\u4e03\u5c42\uff0c\u8fd9\u6b21\u5377\u79ef\u4ed6\u53c8\u505a\u4e86\u4e00\u6b21\u4e0b\u91c7\u6837\uff0c\u4f46\u662f\u5728\u6211\u4eec\u8fd9\u91cc\u7684\u8bdd\uff0c\u7b2c 11 \u5c42\u4ed6\u53c8\u505a\u4e86\u4e00\u6b21\u4e0a\u91c7\u6837\uff0c\u76f8\u4e8e\u90a3\u6211\u7684\u8fd9\u4e2a\u5c31\u4e0a\u91c7\u6837\u4e4b\u540e\uff0c\u548c\u6211\u505a\u4e0b\u91c7\u6837\u4e4b\u524d\u7684\u8fd9\u4e2a\u7ef4\u5ea6\uff0c\u4ed6\u4eec\u4e24\u4e2a\u662f\u4e00\u6837\u7684\uff0c\u6240\u4ee5\u8bf4\u6211\u4eec\u8fd8\u53ef\u4ee5\u628a\u5b83\u62fc\u8d77\u6765\uff0c\u7136\u540e\u7d27\u63a5\u7740\u518d\u8fc7\u4e00\u5c42\u8fd9\u4e2a C3\uff0c\u7136\u540e\u5230\u5b9e\u9645 13 \u5c42\u4e0b\u9762\u7684\u8bdd\u7ee7\u7eed\u5f80\u4e0b\u8d70\uff0c\u7136\u540e\u53c8\u7ecf\u8fc7\u4e00\u5c42\u5377\u79ef\uff0c\u90a3\u90a3\u8fd9\u91cc\u7684\u5377\u79ef\u6211\u4eec\u770b\u4e00\u773c\uff0c\u5c31\u4ed6\u8fd9\u91cc\u7684\u5377\u79ef\uff0c\u5176\u5b9e\u4f60\u4f1a\u53d1\u73b0\u5b83\u5e76\u6ca1\u6709\u518d\u53bb\u505a\u91c7\u6837\uff0c\u5c31\u662f\u8bf4\u5728\u505a\u4e0b\u91c7\u6837\uff0c\u800c\u662f\u8ba9\u5b83\u7ef4\u5ea6\u4fdd\u6301\u4e0d\u53d8\uff0c\u7136\u540e\u6211\u4eec\u5728\u8fd9\u8fd9\u4e2a\u57fa\u7840\u4e0a\u518d\u53bb\u505a\u4e00\u6b21\u4e0a\u91c7\u6837\uff0c\u5c31\u662f\u8bf4\u5728 P7 \u7684\u65f6\u5019\u8fd8\u53d8\u6210\u4e86 2020\uff0c\u90a3 20\u00d720 \u6211\u4eec\u518d\u505a\u4e00\u6b21\u4e0a\u91c7\u6837\uff0c\u53d8\u6210 40\u00d740\uff0c\u90a3 40\u00d740 \u518d\u505a\u4e0a\u91c7\u6837\u4e4b\u540e\uff0c\u73b0\u5728\u5c31\u4f1a\u53d8\u6210 80\u00d780\uff0c\u5c31\u8fd9\u91cc\u7684\u8fd9\u4e2a\u4e0a\u91c7\u6837\u53d8\u6210 80\u00d780\uff0c\u7136\u540e\u8fd9\u91cc\u7684\u8fd9\u4e2a contact16 \u5c31\u662f\u7b2c\u5c31\u8fd9\u4e2a\u7b2c 16 \u5c42\uff0c\u4ed6\u662f\u628a\u7b2c 15 \u5c42\u7684\u8f93\u51fa\uff0c\u4ee5\u53ca\u7b2c\u56db\u5c42\u7684\u8f93\u51fa\u62fc\u8d77\u6765\u4e86\uff0c\u5c31\u662f\u8bf4\u6211\u4eec\u5f80\u524d\u9762\u53bb\u5012\u8fd9\u4e2a C3 \u5bf9\u5427\uff0c\u7136\u540e\u628a\u5b83\u62fc\u63a5\u8d77\u6765\uff0c\u7136\u540e\u7d27\u63a5\u7740\u7ee7\u7eed\u5f80\u4e0b\u8d70\u5230\uff0c\u7136\u540e\u53c8\u5230\u4e86\u4e00\u5c42\uff0c\u8fd9\u4e2a C3C3 \u5b8c\u4e4b\u540e\uff0c\u4ed6\u53c8\u505a\u4e86\u4e00\u6b21\u5377\u79ef\uff0c\u4f46\u8fd9\u6b21\u5377\u79ef\u7684\u65f6\u5019\u4ed6\u53c8\u505a\u4e86\u4e00\u6b21\u4e0b\u91c7\u6837\uff0c\u7136\u540e\u518d\u62fc\u63a5\uff0c\u7136\u540e\u518d C3\uff0c\u7136\u540e\u518d CNV\uff0c\u7136\u540e\u628a\u5b83\u770b\u5f00\u4e86\u8d77\u6765\uff0c\u7136\u540e\u4ed6\u4e00\u76f4\u4e00\u76f4\u5230 C23\uff0c\u7136\u540e\u5c31\u76f8\u5f53\u4e8e\u6240\u6709\u7684\u524d\u9762\u7684\u8fd9\u4e2a\uff0c\u8fd9\u4e00\u90e8\u5206\u5c31\u90fd\u7ed3\u675f\u4e86\uff0c\u5c31\u4e0d\u7ba1\u5b83\u600e\u4e48\u62fc\u5427\uff0c\u4f46\u662f\u5b83\u7684\u539f\u7406\u662f\u4e00\u6837\u7684\uff0c\u90a3\u6700\u540e\u4e00\u5c42\u5230\u4e86\u8fd9\u4e2a detect \u5c42\uff0c\u5b83\u662f\u628a\u7b2c 17 \u5c42\u7684\u8fd9\u4e2a\u7279\u5f81\u56fe\uff0c\u8fd8\u6709\u7b2c 20 \u5c42\uff0c\u8fd8\u6709\u7b2c 23 \u5c42\u7684\u7279\u5f81\u90fd\u90fd\u62ff\u51fa\u6765\u4e86\uff0c\u90a3\u53ef\u4ee5\u770b\u5230\u4ed6\u8fd9\u91cc\u6709\u4ecb\u7ecd\uff0c\u5c31\u662f\u8bf4\u6211\u7b2c 17 \u5c42\u7684\u8fd9\u4e2a\u8f93\u51fa\uff0c\u5176\u5b9e\u4ed6\u662f\u505a\u4e86\u516b\u500d\u4e0b\u91c7\u6837\uff0c\u90a3\u6574\u4e2a\u7279\u5f81\u56fe\u5c31\u53d8\u6210\u4e86 80\u00d780\uff0c\u7136\u540e\u8fd9\u4e2a\u7684\u8bdd\u662f\u505a\u4e86 16 \u500d\u4e0b\u91c7\u6837\uff0c\u8fd9\u4e2a\u7b2c\u4e8c\u4e09\u5c42\u7684\u8bdd\uff0c\u4ed6\u662f\u505a\u4e86 32 \u500d\u4e0b\u91c7\u6837\uff0c\u90a3\u540e\u9762\u8fd9\u4e2a small \u548c\u8fd9\u4e2a media\uff0c\u8fd8\u6709\u8fd9\u4e2a large \u662f\u5e72\u4ec0\u4e48\u7684\uff0c\u662f\u4ec0\u4e48\u610f\u601d\u5462\uff0c\u5c31\u662f\u8bf4\u6211\u4eec\u53ef\u4ee5\u7528\u7528\u8fd9\u4e00\u5c42\uff0c\u5c31\u662f\u8bf4\u7b2c 17 \u5c42\u8fd9\u4e2a 800 \u4e0b\u91c7\u6837\u7684\u7279\u5f81\u56fe\uff0c\u662f\u7528\u6765\u68c0\u6d4b\u5c0f\u76ee\u6807\u7684\uff0c\u7136\u540e\u8fd9\u4e00\u5c42 40\u00d740 \u7684\u8bdd\uff0c\u5b83\u662f\u7528\u6765\u68c0\u6d4b\u8fd9\u4e2a\u4e2d\u7b49\u5927\u5c0f\u7684\u76ee\u6807\u7684\uff0c\u6700\u540e\u4e00\u5c42 32 \u500d\u4e0b\u91c7\u6837\u7684\u8fd9\u4e2a\u7279\u5f81\u56fe\uff0c\u5b83\u662f\u7528\u6765\u68c0\u6d4b\u8fd9\u4e9b\u6bd4\u8f83\u5927\u7684\u7269\u4f53\u7684\u7ed3\u6784\u7684\u8bdd\uff0c\u5927\u6982\u5c31\u662f\u8fd9\u6837\u7684\u4e00\u4e2a\u7ed3\u6784\uff0c\u90a3\u6211\u4eec\u5176\u5b9e\u540e\u7eed\u8981\u4fee\u6539\u8fd9\u4e2a\u6a21\u578b\u7684\u8bdd\uff0c\u5176\u5b9e\u5c31\u662f\u5728\u8fd9\u91cc\u9762\u4fee\u6539\u4e86\u3002<\/p>\n\n\n\n
YOLOv5s \u7f51\u7edc\u7ed3\u6784\u6982\u89c8<\/h2>\n\n\n\n
\u8f93\u5165\u5c42:
\u8f93\u5165\u5c3a\u5bf8: 1x3x640x640 (\u6279\u91cf\u5927\u5c0f x \u901a\u9053\u6570 x \u9ad8\u5ea6 x \u5bbd\u5ea6)
Backbone (\u4e3b\u5e72\u7f51\u7edc):<\/p>\n\n\n\n
Conv0: \u521d\u59cb\u5377\u79ef\u5c42, \u5c06\u8f93\u5165\u4ece 3x640x640 \u8f6c\u6362\u4e3a 32x320x320\u3002\u6b64\u6b65\u9aa4\u8fdb\u884c\u4e861\/2\u7684\u4e0b\u91c7\u6837\u3002
C3\u6a21\u5757:
C3_1: \u8f93\u5165 32x320x320 -> \u8f93\u51fa 64x160x160 (1\/4\u4e0b\u91c7\u6837)
C3_2: \u8f93\u5165 64x160x160 -> \u8f93\u51fa 128x80x80 (1\/8\u4e0b\u91c7\u6837)
C3_3: \u8f93\u5165 128x80x80 -> \u8f93\u51fa 256x40x40 (1\/16\u4e0b\u91c7\u6837)
C3_4: \u8f93\u5165 256x40x40 -> \u8f93\u51fa 512x20x20 (1\/32\u4e0b\u91c7\u6837)
SPPF (Spatial Pyramid Pooling - Fast): \u5728\u6700\u540e\u4e00\u4e2aC3\u6a21\u5757\u540e\u5e94\u7528\uff0c\u589e\u5f3a\u6a21\u578b\u5bf9\u4e0d\u540c\u5c3a\u5ea6\u7279\u5f81\u7684\u6355\u6349\u80fd\u529b\u3002
Neck (\u9888\u90e8\u7f51\u7edc):<\/p>\n\n\n\n
\u4e0a\u91c7\u6837\u4e0e\u62fc\u63a5:
\u7b2c\u4e00\u6b21\u4e0a\u91c7\u6837: 512x20x20 -> 256x40x40\uff0c\u4e0eC3_3\u7684\u8f93\u51fa\u62fc\u63a5\uff0c\u5f62\u6210 512x40x40\u3002
\u7b2c\u4e8c\u6b21\u4e0a\u91c7\u6837: 256x40x40 -> 128x80x80\uff0c\u4e0eC3_2\u7684\u8f93\u51fa\u62fc\u63a5\uff0c\u5f62\u6210 256x80x80\u3002
C3\u6a21\u5757:
C3_5: \u8f93\u5165 512x40x40 -> \u8f93\u51fa 256x40x40
C3_6: \u8f93\u5165 256x80x80 -> \u8f93\u51fa 128x80x80
Head (\u5934\u90e8\u7f51\u7edc):<\/p>\n\n\n\n
\u68c0\u6d4b\u5c42:
\u7b2c17\u5c42: \u8f93\u5165 256x40x40 -> \u8f93\u51fa 255x40x40 (\u7528\u4e8e\u68c0\u6d4b\u4e2d\u7b49\u5927\u5c0f\u76ee\u6807)
\u7b2c20\u5c42: \u8f93\u5165 128x80x80 -> \u8f93\u51fa 255x80x80 (\u7528\u4e8e\u68c0\u6d4b\u5c0f\u76ee\u6807)
\u7b2c23\u5c42: \u8f93\u5165 512x20x20 -> \u8f93\u51fa 255x20x20 (\u7528\u4e8e\u68c0\u6d4b\u5927\u76ee\u6807)
\u6bcf\u4e2a\u9636\u6bb5\u7684\u8f93\u51fa\u5c3a\u5bf8\u53cd\u6620\u4e86\u4e0d\u540c\u5c3a\u5ea6\u4e0b\u7684\u7279\u5f81\u56fe\uff0c\u8fd9\u4e9b\u7279\u5f81\u56fe\u88ab\u7528\u4e8e\u591a\u5c3a\u5ea6\u76ee\u6807\u68c0\u6d4b\uff0c\u4ece\u800c\u63d0\u9ad8\u4e86\u6a21\u578b\u5bf9\u4e8e\u4e0d\u540c\u5927\u5c0f\u76ee\u6807\u7684\u68c0\u6d4b\u80fd\u529b\u3002\u901a\u8fc7\u4e0a\u91c7\u6837\u548c\u7279\u5f81\u56fe\u62fc\u63a5\uff0cYOLOv5s\u80fd\u591f\u6709\u6548\u5730\u5229\u7528\u6765\u81ea\u591a\u4e2a\u5c3a\u5ea6\u7684\u4fe1\u606f\uff0c\u8fd9\u5bf9\u4e8e\u63d0\u5347\u68c0\u6d4b\u7cbe\u5ea6\u81f3\u5173\u91cd\u8981\u3002<\/p>\n\n\n\n
\u6211\u4eec\u4ecetrain.py\u770b\u8d77\uff0c\u76f4\u63a5\u770bModel\uff0c\u7b2c132\u884c\u5c31\u662f\u521b\u5efayaml\u7684\u6587\u4ef6\u4ee3\u7801\uff0c\u770b\u5230\u521b\u5efa\u4e86yolo.py\u91cc\u9762\u7684DetectionModel\uff0c\u4e0b\u65b9\u6709178\u884c\u548cnc\u3002<\/p>\n\n\n\n
\u7406\u8bba\u5199\u8d77\u6765\u592a\u9ebb\u70e6\u5566\uff01               \u9119\u4eba\u6ca1\u6709\u57fa\u7840\u5f88\u96be\u5168\u5199\u51fa\u6765\uff01\u6b64\u5904\u6682\u672a\u5b8c\u5de5\uff01\uff01\uff01<\/strong><\/p>\n\n\n\n
\u4fee\u6539\u7f51\u7edc\u7ed3\u6784-\u4ee5C2f\u4e3a\u4f8b<\/h1>\n\n\n\n
\u6211\u4eec\u501f\u9274v8\u7684\u4ee3\u7801https:\/\/github.com\/ultralytics\/ultralytics (YOLOv8)<\/p>\n\n\n\n
\u4fee\u6539\u987a\u5e8f\u5982\u4e0b
models\/commons.py------>\u52a0\u5165\u65b0\u589e\u7f51\u7edc\u7ed3\u6784
models\/yolo.py\u2014---------->\u8bbe\u5b9a\u7f51\u7edc\u7ed3\u6784\u7684\u4f20\u53c2\u7ec6\u8282
models\/yolov5*.yaml----->\u4fee\u6539\u73b0\u6709\u6a21\u578b\u7ed3\u6784\u914d\u7f6e\u6587\u4ef6\u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 train.py---------------------->\u8bad\u7ec3\u65f6\u6307\u5b9a\u6a21\u578b\u7ed3\u6784\u914d\u7f6e\u6587\u4ef6<\/p>\n\n\n\n
\u6253\u5f00v8\u4e0b\u7684nn\u7f51\u7edc\u7ed3\u6784\u6587\u4ef6\u5939\u3002\u6211\u4eec\u6ce8\u610f\u5230yolov5\u91cc\u9762\u5728common.py\u4e2d\u5199\u4e86\u8fd9\u4e2a\u7ed3\u6784\u7684\u5b9a\u4e49\uff0c\u5728v8\u4e2d\uff0c\u5168\u90e8\u5199\u5728\u4e86module\u4e2d<\/p>\n\n\n\n
\u6211\u4eec\u53ef\u4ee5\u628amodules.py\u7684C2f\u5b8c\u5168\u590d\u5236\u5230commons.py\u7684C3\u524d\u9762\uff1a\u5728\u6700\u65b0\u72482023.8\u4e2d\uff0cmodules.py\u4ee5\u53ca\u53d8\u6210\u4e86\u5305\u542b\u591a\u4e2a\u6587\u4ef6\u7684\u6587\u4ef6\u5939\uff0c\u6253\u5f00block.py\u627e\u5230C2f<\/strong><\/p>\n\n\n\n
class C2f(nn.Module):\n    \"\"\"Faster Implementation of CSP Bottleneck with 2 convolutions.\"\"\"\n\n    def __init__(self, c1, c2, n=1, shortcut=False, g=1, e=0.5):\n        \"\"\"Initializes a CSP bottleneck with 2 convolutions and n Bottleneck blocks for faster processing.\"\"\"\n        super().__init__()\n        self.c = int(c2 * e)  # hidden channels\n        self.cv1 = Conv(c1, 2 * self.c, 1, 1)\n        self.cv2 = Conv((2 + n) * self.c, c2, 1)  # optional act=FReLU(c2)\n        self.m = nn.ModuleList(Bottleneck(self.c, self.c, shortcut, g, k=((3, 3), (3, 3)), e=1.0) for _ in range(n))\n\n    def forward(self, x):\n        \"\"\"Forward pass through C2f layer.\"\"\"\n        y = list(self.cv1(x).chunk(2, 1))\n        y.extend(m(y[-1]) for m in self.m)\n        return self.cv2(torch.cat(y, 1))\n\n    def forward_split(self, x):\n        \"\"\"Forward pass using split() instead of chunk().\"\"\"\n        y = list(self.cv1(x).split((self.c, self.c), 1))\n        y.extend(m(y[-1]) for m in self.m)\n        return self.cv2(torch.cat(y, 1))<\/code><\/pre>\n\n\n\n
\u5176\u4e2d \u7531\u4e8ek\u53c2\u6570\u672a\u5b9a\u4e49\uff0c\u6211\u4eec\u9700\u8981copy\u5b9a\u4e49\u5730\u65b9Bottleneck\uff0c\u6539\u540d\u4e3aC2f...\uff0c\u540c\u65f6C2f\u4e2d\u4e5f\u8981\u6539<\/p>\n\n\n\n
class C2fBottleneck(nn.Module):\n    \"\"\"Standard bottleneck.\"\"\"\n\n    def __init__(self, c1, c2, shortcut=True, g=1, k=(3, 3), e=0.5):\n        \"\"\"Initializes a standard bottleneck module with optional shortcut connection and configurable parameters.\"\"\"\n        super().__init__()\n        c_ = int(c2 * e)  # hidden channels\n        self.cv1 = Conv(c1, c_, k[0], 1)\n        self.cv2 = Conv(c_, c2, k[1], 1, g=g)\n        self.add = shortcut and c1 == c2\n\n    def forward(self, x):\n        \"\"\"Applies the YOLO FPN to input data.\"\"\"\n        return x + self.cv2(self.cv1(x)) if self.add else self.cv2(self.cv1(x))<\/code><\/pre>\n\n\n\n
\u90a3\u4e48\u5728\u539f\u5148\u4f20\u5165C3\u7684\u5730\u65b9\u52a0\u5165C2f\u5c31\u597d\u4e86\uff1a\u5728yolo.py parse_model\u4e0b319\u884c\u52a0C2f\uff0c325\u884c\u540c\u7406<\/p>\n\n\n\n
if m in {BottleneckCSP, C3, C3TR, C3Ghost, C3x,C2f}:<\/code><\/pre>\n\n\n\n
\u63a5\u4e0b\u6765\u4fee\u6539 yolov5s.yaml,\u628abackbone\u4e2d\u7684C3\u5168\u6539\u6210C2f,head\u90e8\u5206\u53ef\u6539\u53ef\u4e0d\u6539<\/p>\n\n\n\n
\u4fee\u6539train.py\u7684435\u884c,\u53ef\u4ee5\u628ayolov5s.yaml\u6539\u4e3a\u81ea\u5b9a\u4e49\u540d\u5b57<\/p>\n\n\n\n
def parse_opt(known=False):\n    parser = argparse.ArgumentParser()\n    parser.add_argument('--weights', type=str, default=ROOT \/ 'yolov5s.pt', help='initial weights path')\n    parser.add_argument('--cfg', type=str, default=ROOT \/ 'models\/yolov5s.yaml', help='model.yaml path')<\/code><\/pre>\n\n\n\n
\u76f4\u63a5python  train.py\u5373\u53ef<\/p>\n\n\n\n
\u53ef\u80fd\u4f1a\u597d\u5947,\u6211\u4eec\u8fd9\u91cc\u4f20\u4e86CONFIG,\u662f\u8fd9\u4e2acf
,\u4f46\u662f\u6211\u4eec\u8fd9\u4e2a\u4f4d\u7f6e\u4f20\u7684yolov5s.pt,\u4ed6\u8fd9\u4e24\u4e2a\u7ed3\u6784\u4e5f\u4e0d\u4e00\u6837,\u8fd8\u80fd\u4e0d\u80fd\u5c31\u662f\u8bf4\u9884\u8bad\u7ec3,\u6216\u8005\u8bf4\u501f\u7528\u4e4b\u524d\u7684\u6743\u91cd?<\/p>\n\n\n\n
\u5176\u5b9e\u662f\u53ef\u4ee5\u7684,\u6211\u4eec\u53ef\u4ee5\u5148\u7b80\u5355\u6765\u770b\u4e00\u4e0b,\u8fd9\u91cc\u5176\u5b9e\u8fd9\u4e2a\u53c2\u6570\u4e5f\u6bd4\u8f83\u5173\u952e,\u5c31\u662f\u8bf4\u5f53\u6211\u4eec\u4f20\u8fd9\u4e2a\u4f4d\u7f6e\u7684\u65f6\u5019,\u4ed6\u5c31\u4f1a\u53bb\u5c1d\u8bd5\u4ece\u8fd9\u4e2a\u6743\u91cd\u91cc\u9762\u8fdb\u884c\u52a0\u8f7d\u6743\u91cd,\u90a3\u5982\u679c\u8bf4\u6211\u4eec\u4e0d\u4f20,\u90a3\u5b83\u5c31\u76f8\u5f53\u4e8e\u5c31\u662f\u5b8c\u5168\u7684\u4ece\u5934\u5f00\u59cb\u8bad\u7ec3\u8fd9\u90e8\u5206\u7684\u903b\u8f91,\u5c31\u662f\u6211\u4eec\u4e0a\u6b21\u8bfe\u8bb2\u7684\u90a3\u4e2a\u6a21\u578b\u521b\u5efa\u7684\u4e00\u90e8\u5206<\/p>\n\n\n\n
\u4ed6\u8fd9\u91cc\u6709\u4e00\u4e2a\u53c2\u6570,\u6709\u4e00\u4e2a\u5224\u65ad\u53eb\u8fd9\u4e2aPRETRAIN,\u5982\u679cpt\u7ed3\u5c3e\u7684\u8bdd,\u8bf4\u660e\u662f\u4e00\u4e2a\u9884\u8bad\u7ec3\u6743\u91cd,\u7136\u540e\u5c31\u4f1a\u5c1d\u8bd5\u53bb\u52a0\u8f7d\u7136\u540e\u628a\u5b83\u7ed9\u62f7\u8d1d\u8fc7\u53bb,\u4f1a\u505a\u4e00\u4e0b\u8fc1\u79fb,\u90a3\u5982\u679c\u8bf4\u4f60\u4e0d\u4f20\u5c31\u6ca1\u6709\u4e86 <\/p>\n\n\n\n
\u5f15\u5165\u6ce8\u610f\u529b\u673a\u5236-\u4ee5SE\u4e3a\u4f8b<\/h1>\n\n\n\n
ZhugeKongan\/Attention-mechanism-implementation: Self-attention\u3001Non-local\u3001SE\u3001SK\u3001CBAM\u3001DANet (github.com)<\/a><\/p>\n\n\n\n
\u5176\u4e2dSE-block.py\u5c31\u662f\u4ee3\u7801\u90e8\u5206\u4e86,\u628a\u4ed6\u653e\u5230common.py \u7684C2f\u524d,\u5e76\u4e14import<\/p>\n\n\n\n
import torch.nn as nn\nimport torch.nn.functional as F\n\nclass SE(nn.Module):\n\n    def __init__(self, in_chnls, ratio):\n        super(SE, self).__init__()\n        self.squeeze = nn.AdaptiveAvgPool2d((1, 1))\n        self.compress = nn.Conv2d(in_chnls, in_chnls \/\/ ratio, 1, 1, 0)\n        self.excitation = nn.Conv2d(in_chnls \/\/ ratio, in_chnls, 1, 1, 0)\n\n    def forward(self, x):\n        out = self.squeeze(x)\n        out = self.compress(out)\n        out = F.relu(out)\n        out = self.excitation(out)\n        return x*F.sigmoid(out)<\/code><\/pre>\n\n\n\n
\u8fd9\u4e2aSE\u5b83\u5230\u5e95\u662f\u505a\u4e86\u4ec0\u4e48? \u200b<\/p>\n\n\n\n
\u5b83\u91cc\u9762\u5b9a\u4e49\u4e86\u4e09\u4e2a\u6a21\u5757\uff0c\u4e00\u4e2a\u662f\u9002\u5e94\u7684\u90a3\u79cd\u5e73\u5747\u503c\u5316\uff0c\u4e00\u4e2a\u662f\u7136\u540e\u518d\u52a0\u4e0a\u4e24\u4e2a\u5377\u79ef\uff0c\u4e00\u4e2a\u662f\u628a\u8fd9\u5c31\u662f\u8f93\u5165\u901a\u9053\u6570\uff0c\u7136\u540e\u5377\u79ef\u4e4b\u540e\u628a\u5b83\u538b\u7f29\u4e00\u70b9\uff0c\u538b\u7f29\u5b8c\u4e4b\u540e\uff0c\u7136\u540e\u518d\u628a\u538b\u7f29\u518d\u8fd4\u56de\uff0c\u518d\u5377\u6210\u4e4b\u524d\u7684\u8f93\u5165\u901a\u9053\u6570\uff0c\u90a3\u5177\u4f53\u505a\u7684\u8bdd\u5c31\u662f\u5148\u8fdb\u884c\u8fd9\u4e2a\u76f4\u753b\uff0c\u7136\u540e\u8fdb\u884c\u4e00\u6b21\u538b\u7f29\uff0c\u7136\u540e\u518d\u8fdb\u884c\u4e00\u6b21\u5c31\u662f\u600e\u4e48\u8bf4\u5462\u518d\u6da8\u56de\u6765\uff0c\u7136\u540e\u7d27\u63a5\u7740\u6211\u4eec\u628a\u6da8\u56de\u6765\u7684\u90a3\u4e2a\u7ed3\u679c\uff0c\u8fdb\u884c\u4e00\u4e2a\u7528\u7528\u90a3\u4e2a sigmoid \u53c2\u6570\u8fdb\u884c\u6fc0\u6d3b\uff0c\u7136\u540e\u518d\u53bb\u4e58\u4e0a\u539f\u59cb\u8fd9\u4e2a x\uff0c\u5c31\u662f\u5b83\u7684\u8f93\u5165\u901a\u9053\uff0c\u8f93\u5165\u901a\u9053\u4ee5\u53ca\u5b83\u7684\u8fd9\u4e2a\u53c2\u6570\uff0c\u90a3\u53ef\u4ee5\u770b\u5230\u5728\u8fd9\u4e2a SE \u7684\u8fd9\u4e2a\u7ed3\u6784\u91cc\u9762\uff0c\u5176\u5b9e\u5b83\u53ea\u6709\u4e24\u4e2a\u53c2\u6570\uff0c\u4e00\u4e2a\u662f in channels\uff0c\u4e00\u4e2a\u662f\u8fd9\u4e2a ratio\uff0cin channelse \u7684\u8bdd\u5176\u5b9e\u5b83\u5c31\u662f\u8f93\u5165\u901a\u9053\u4e86\uff0c\u53ea\u4e0d\u8fc7\u5728 SE1 \u8fd9\u4e2a\u91cc\u9762\u5b83\u6bd4\u8f83\u7279\u6b8a\u4e00\u70b9\uff0c\u4ed6\u7684\u8fd9\u4e2a\u8f93\u5165\u901a\u9053\u548c\u8f93\u5165\u901a\u9053\u5176\u5b9e\u662f\u4e00\u6837\u7684\uff0c\u5bf9\u5427\uff0c\u5c31\u662f\u4ed6\u8fd9\u4e2a\u5148\u538b\u7f29\uff0c\u540e\u6765\u53c8\u7ed9\u5b83\u8f6c\u56de\u53bb\u662f\u4e00\u4e2a\u503c\uff0c\u8fd9\u8ddf\u6211\u4eec\u4e0a\u6b21\u8fd9\u4e2a cf \u5b83\u6709\u4e24\u4e2a\u901a\u9053\uff0c\u8fd9\u4e2a\u5176\u5b9e\u662f\u6709\u4e9b\u5dee\u5f02\u7684\u5bf9\u5427\uff0c\u7136\u540e\u63a5\u4e0b\u6765\u7684\u8bdd\u5b83\u8fd8\u6709\u4e2a ritual\uff0c\u8fd9\u4e2a\u503c\u5c31\u662f\u8bf4\u6211\u4e2d\u95f4\u8fd9\u4e2a\u94f6\u5c42\uff0c\u5c31\u662f\u6211\u7b2c\u4e00\u6b21\u5377\u79ef\u4e4b\u540e\uff0c\u4ed6\u7684\u90a3\u4e2a\u8f93\u5165\u901a\u9053\u6570\uff0c\u5e94\u8be5\u662f\u6211\u8fd9\u4e2a in china \u9664\u4ee5\u8fd9\u4e2a ratio\uff0c\u5c31\u662f\u90a3\u4e2a\u8f93\u5165\u901a\u9053\u6570\uff0c\u76f8\u5f53\u4e8e\u6574\u4e2a\u7684\u8fd9\u4e2a SE \u7684\u8fd9\u4e2a\u7ed3\u6784\uff0c\u6211\u4eec\u4e00\u5171\u9700\u8981\u4f20\u9012\u4e24\u4e2a\u53c2\u6570\u5bf9\u5427\uff0cOK \u6211\u4eec\u8981\u628a\u8fd9\u4e2a\u8bb0\u6e05\u695a\u3002<\/p>\n\n\n\n
\u200b<\/p>\n\n\n\n
\u63a5\u4e0b\u6765\u5728yolov5s.yaml\u4e2dbackbone\u6700\u540e\u4e00\u884c\u6dfb\u52a0<\/p>\n\n\n\n
[-1,1,SE,[1024,2]]<\/code><\/pre>\n\n\n\n
<\/p>\n\n\n\n
\u90a3\u4e48\u8fd9\u4e00\u5c42\u53d8\u6210\u7b2c10\u5c42,\u63a5\u4e0b\u6765head\u6bcf\u4e00\u884c\u90fd\u8981\u52a0\u4e00\u5c42 ,\u5176\u4e2d\u5927\u4e8e10\u7684\u90fd\u8981\u8fdb\u884c\u4fee\u6539+1<\/p>\n\n\n\n
head:\n  [[-1, 1, Conv, [512, 1, 1]],\n   [-1, 1, nn.Upsample, [None, 2, 'nearest']],\n   [[-1, 6], 1, Concat, [1]],  # cat backbone P4\n   [-1, 3, C3, [512, False]],  # 13\n\n   [-1, 1, Conv, [256, 1, 1]],\n   [-1, 1, nn.Upsample, [None, 2, 'nearest']],\n   [[-1, 4], 1, Concat, [1]],  # cat backbone P3\n   [-1, 3, C3, [256, False]],  # 17 (P3\/8-small)\n\n   [-1, 1, Conv, [256, 3, 2]],\n   [[-1, 15], 1, Concat, [1]],  # cat head P4\n   [-1, 3, C3, [512, False]],  # 20 (P4\/16-medium)\n\n   [-1, 1, Conv, [512, 3, 2]],\n   [[-1, 11], 1, Concat, [1]],  # cat head P5\n   [-1, 3, C3, [1024, False]],  # 23 (P5\/32-large)\n\n   [[18, 21, 24], 1, Detect, [nc, anchors]],  # Detect(P3, P4, P5)\n  ]\n <\/code><\/pre>\n\n\n\n
<\/p>\n\n\n\n
 \u5728yolo.py\u4e2d\u6dfb\u52a0343\u884c<\/p>\n\n\n\n
elif m is SE:\n    c1= ch[f]#\u8fd9\u91cc\u8981\u6ce8\u610f,\u4e3a\u4ec0\u4e48\u8981\u8bbe\u5b9aC2,\u800c\u4e0d\u662f\u7528C1,\u56e0\u4e3aparse_model\u4e2d,c2\u662f\u7528\u6765\u8bb0\u5f55\u8f93\u51fa\u7ef4\u5ea6\u7684,\u6700\u540e\u4f1a\u5c06c2\u5b58\u5165ch\u4e2d,\u6240\u4ee5\u4e00\u5b9a\u5bf9c2\u7684\u503c\u8fdb\u884c\u8bbe\u5b9a\n    c2= args[0]\n    if c2 \uff01= no: # if not output\n        c2 = make_divisible(c2 * gw, 8)\n    args =[c1, args[1]\n<\/code><\/pre>\n\n\n\n
<\/p>\n\n\n\n
\u66ff\u6362\u4e3b\u5e72\u7f51\u7edc-\u4ee5MobileNet\u4e3a\u4f8b<\/h1>\n\n\n\n
\u65b0\u5efademo.ipynb\u6587\u4ef6\u5e76\u4e14pip install torchinfo<\/p>\n\n\n\n
\u6b64\u5904\u7f3a\u5c11\u90e8\u5206\u5b8c\u5584<\/strong><\/p>\n\n\n\n
\u5728\u66ff\u6362\u4e3b\u5e72\u7f51\u7edc\u7684\u90e8\u5206\u9700\u8981\u8003\u8651\u7684\u4e00\u70b9<\/h3>\n\n\n\n
\"\"\u200b\u7f16\u8f91<\/p>\n\n\n\n
\u5176\u5b9e\u662f\u5f88\u6709\u8003\u7a76\u7684\u5bf9\u5427
\u4e00\u5b9a\u8981\u6709\u5b83\u7ecf\u8fc7\u516b\u500d\u91c7\u6837\u4ee5\u53ca16\u500d\u91c7\u683732\u500d\u7684\u91c7\u6837\u7684\u8f93\u51fa<\/strong>
\u4e3a\u4ec0\u4e48\u5462\uff1f\u56e0\u4e3a\u9996\u5148\u554a\u4f60\u7684\u8fd9\u4e2a\u516b\u500d\u7684\u91c7\u6837\u4f1a\u5728\u8fd9\u513f\u6765\u7528
\u4f1a\u5728\u5c31\u662f\u7b2c17\u5c42\u524d\u9762\u6765\u7528
\u7136\u540e\u4ee516\u500d\u7684\u91c7\u6837\u4f1a\u5728head\u7684\u7b2c\u4e00\u90e8\u5206\u6765\u7528
\u7136\u540e\u4f60\u8981\u60f3\u7528\u5b83\u7684\u8bdd
\u4f60\u9996\u5148\u4f60\u4fdd\u8bc1\u4f60\u7684\u8fd9\u4e2a\u8fdbhead\u4e4b\u524d
\u5b83\u4e00\u5b9a\u662f\u4e00\u4e2a20\u00d720
\u5c31\u662f32\u500d\u4e0b\u91c7\u6837\u7684\u4e00\u4e2a\u7ed3\u6784<\/s>
\u6211\u4eec\u7684\u4e3b\u5e72\u7f51\u7edc\u66ff\u6362\u5b8c\u6210\u4e4b\u540e,\u4e00\u5b9a\u4e5f\u662f\u6709\u7ecf\u8fc7\u8fd9\u51e0\u6b21\u91c7\u6837\u7684,\u4e00\u5b9a\u8981\u6709\u4e00\u5c42,\u5b83\u7684\u90a3\u4e2a\u7ed3\u6784\u5bf9\u5e94\u7684\u662f\u516b\u500d\u4e0b\u91c7\u6837,\u4e00\u5b9a\u8981\u6709\u4e00\u5c42\u5bf9\u5e94\u7684\u662f16\u500d\u4e0b\u91c7\u6837,\u8fd8\u6709\u4e00\u5c42\u8981\u5bf9\u5e94\u7684\u662f32\u500d\u4e0b\u91c7\u6837
\u53ea\u6709\u8fd9\u6837\u7684\u8bdd\u624d\u80fd\u5b8c\u7f8e\u7684\u53bb\u7ee7\u627f\u6574\u4e2a\u7ed3\u6784,\u53ef\u4ee5\u8ba9\u8fd9\u4e2ahigh\u7684\u90e8\u5206\u76f4\u63a5\u5c31\u53bb\u590d\u7528.<\/p>\n\n\n\n
import torchvision.models as models\nfrom torchinfo import summary\nmodel = models.mobilenet_v3_small(pretrained=True, progress=True)\nsummary(model.features, input_size=(1, 3, 640, 640))\nmodel\ntype(model.features)\nmodel.features[:4]\nmodel.features[4:9]\nmodel.features[9:]\nimport fiftyone as fo\nimport fiftyone.zoo as foz\ndataset = foz.load_zoo_dataset(\n    \"open-images-v7\",\n    split=\"validation\",\n    label_types=[\"detections\"],\n    classes=[\"Mouse\"],\n    max_samples=100,\n    seed=51,\n    shuffle=True,\n    dataset_name=\"mouse\"\n)\nfo.launch_app(dataset)<\/code><\/pre>\n\n\n\n
\u5728common.py\u4e2d\u6dfb\u52a0\u5982\u4e0b<\/p>\n\n\n\n
class MobileNetV3(nn.Module):\n\n    def init(self, slice):\n        super(MobileNetV3, self). init()\n        self.model =None\n        if slice == 1\uff1a\n            self.model= models.mobilenet_v3_small(pretrained=True).features[:4]\n        elif slice == 2\uff1a\n            self.model=models.mobilenet v3 small(pretrained=True).features[4:9]\n        else:\n            self.model=models.mobilenet_v3_small(pretrained=True).features[9:]\n\n    def forward(self, x):\n        return self.model(x)<\/code><\/pre>\n\n\n\n
\u4fee\u6539yaml
 \u200b\"\"\"\"<\/p>\n\n\n\n
\u00a0yolo.py\u6dfb\u52a0Mobile\u7684elif\uff08\u5982\u4e0b\uff09\uff0c\u5b9e\u73b0\u4e86\u6a21\u578b\u8f7b\u91cf\u5316\u7684\u5de5\u4f5c ---------\u6570\u636e702w\u5230373w\u7684\u8f7b\u91cf\u5316<\/p>\n\n\n\n
elif m is SE:\n    c1 = ch[f]\n    c2 = args[0]\n    if c2 \uff01= no: # if not output\n        c2 = make_divisible(c2 * gw, 8)\n    args = [c1, args[1]]\nelif m is MobileNetV3:\n    c2 = args[0]\n    args= args[1:]\nelse:\n    c2= ch[f]<\/code><\/pre>\n","protected":false},"excerpt":{"rendered":"
AutoDL\u670d\u52a1\u5668\u90e8\u7f72\u8bad\u7ec3 \u6ce8\u518c\u8fc7\u7a0b\u4e0d\u4e88\u663e\u793a\uff0c\u7b80\u5355\u7684yolov5\u5176\u5b9e\u4e0d\u9700\u8981\u4ec0\u4e48\u9ad8\u7ea7\u5361\uff0c\u6bd5\u7adf\u8fdc\u90531050-4G\u7248\uff0c\u8fd1\u52304060-8g …<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"emotion":"","emotion_color":"","title_style":"","license":"","footnotes":""},"categories":[3],"tags":[6,5],"class_list":["post-273","post","type-post","status-publish","format-standard","hentry","category-3","tag-cv","tag-yolo"],"_links":{"self":[{"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/posts\/273","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/eve2333.top\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=273"}],"version-history":[{"count":0,"href":"https:\/\/eve2333.top\/index.php?rest_route=\/wp\/v2\/posts\/273\/revisions"}],"wp:attachment":[{"href":"https:\/\/eve2333.top\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=273"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/eve2333.top\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=273"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/eve2333.top\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=273"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}