{"id":6569,"date":"2024-09-25T19:03:24","date_gmt":"2024-09-25T11:03:24","guid":{"rendered":"https:\/\/www.aisharenet.com\/?p=6569"},"modified":"2025-05-13T22:10:23","modified_gmt":"2025-05-13T14:10:23","slug":"codeformer","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/en\/codeformer\/","title":{"rendered":"CodeFormer\uff1a\u56fe\u50cf\u4e0e\u89c6\u9891\u9762\u90e8\u590d\u539f\uff0c\u8001\u7167\u7247\u4fee\u590d\uff0c\u63d0\u4f9b\u4e00\u952e\u90e8\u7f72\u7248"},"content":{"rendered":"

CodeFormer \u7efc\u5408\u4ecb\u7ecd<\/h2>\n

CodeFormer \u662f\u4e00\u4e2a\u9762\u5411\u9c81\u68d2\u76f2\u8138\u4fee\u590d\u7684\u4ee3\u7801\u5e93\uff0c\u7531\u5357\u6d0b\u7406\u5de5\u5927\u5b66 S-Lab \u7684\u7814\u7a76\u56e2\u961f\u5f00\u53d1\uff0c\u5e76\u5728 NeurIPS 2022 \u4e0a\u53d1\u8868\u3002\u8be5\u9879\u76ee\u5229\u7528\u4ee3\u7801\u672c\u67e5\u627e\u53d8\u6362\u5668\uff08Codebook Lookup Transformer\uff09\u6280\u672f\uff0c\u65e8\u5728\u63d0\u5347\u56fe\u50cf\u4e2d\u4eba\u8138\u7684\u4fee\u590d\u6548\u679c\uff0c\u7279\u522b\u662f\u5728\u5904\u7406\u4f4e\u8d28\u91cf\u6216\u635f\u574f\u7684\u56fe\u50cf\u65f6\u8868\u73b0\u51fa\u8272\u3002CodeFormer \u63d0\u4f9b\u4e86\u591a\u79cd\u529f\u80fd\uff0c\u5305\u62ec\u4eba\u8138\u4fee\u590d\u3001\u4e0a\u8272\u548c\u4fee\u8865\u7b49\uff0c\u9002\u7528\u4e8e\u5404\u79cd\u56fe\u50cf\u5904\u7406\u9700\u6c42\u3002\u6b64\u5916\uff0c\u9879\u76ee\u652f\u6301\u89c6\u9891\u8f93\u5165\uff0c\u5e76\u63d0\u4f9b\u6613\u4e8e\u4f7f\u7528\u7684\u5728\u7ebf\u6f14\u793a\u3001\u9884\u8bad\u7ec3\u6a21\u578b\u548c\u8be6\u7ec6\u7684\u4f7f\u7528\u8bf4\u660e\u3002<\/p>\n

\u5fc5\u987b\u4e25\u683c\u9605\u8bfb\u4f7f\u7528\u8bf4\u660e\uff0c\u5426\u5219\u65e0\u6cd5\u6b63\u5e38\u8fd0\u884c\u3002\u6b63\u5f0f\u4f5c\u4e3a\u8001\u7167\u7247\u4fee\u590d\u5546\u7528\u552e\u5356\u65f6\uff0c\u9700\u8981\u7b80\u5355PS\u57fa\u7840\uff0c\u5426\u5219\u8fbe\u4e0d\u5230\u7f51\u4e0a\u540c\u7b49\u6548\u679c\u3002<\/p><\/blockquote>\n

 <\/p>\n

\"CodeFormer\uff1a\u56fe\u50cf\u4e0e\u89c6\u9891\u9762\u90e8\u590d\u539f|\u9762\u90e8\u6062\u590d|\u8001\u7167\u7247\u4fee\u590d,\u63d0\u4f9b\u4e00\u952e\u90e8\u7f72-1\"<\/p>\n

 <\/p>\n

\"CodeFormer\uff1a\u56fe\u50cf\u4e0e\u89c6\u9891\u9762\u90e8\u590d\u539f|\u9762\u90e8\u6062\u590d|\u8001\u7167\u7247\u4fee\u590d,\u63d0\u4f9b\u4e00\u952e\u90e8\u7f72-1\"

CodeFormer \u5c1d\u8bd5\u589e\u5f3a\u65e7\u7167\u7247\/\u4fee\u590dAI\u4eba\u50cf<\/p><\/div>\n

 <\/p>\n

\"CodeFormer\uff1a\u56fe\u50cf\u4e0e\u89c6\u9891\u9762\u90e8\u590d\u539f|\u9762\u90e8\u6062\u590d|\u8001\u7167\u7247\u4fee\u590d,\u63d0\u4f9b\u4e00\u952e\u90e8\u7f72-2\"

CodeFormer \u9762\u90e8\u4fee\u590d<\/p><\/div>\n

 <\/p>\n

\"CodeFormer\uff1a\u56fe\u50cf\u4e0e\u89c6\u9891\u9762\u90e8\u590d\u539f|\u9762\u90e8\u6062\u590d|\u8001\u7167\u7247\u4fee\u590d,\u63d0\u4f9b\u4e00\u952e\u90e8\u7f72-3\"

CodeFormer \u9762\u90e8\u8272\u5f69\u589e\u5f3a\u548c\u6062\u590d<\/p><\/div>\n

 <\/p>\n

\"CodeFormer\uff1a\u56fe\u50cf\u4e0e\u89c6\u9891\u9762\u90e8\u590d\u539f|\u9762\u90e8\u6062\u590d|\u8001\u7167\u7247\u4fee\u590d,\u63d0\u4f9b\u4e00\u952e\u90e8\u7f72-4\"

CodeFormer \u8138\u90e8\u4fee\u590d<\/p><\/div>\n

 <\/p>\n

 <\/p>\n

CodeFormer \u529f\u80fd\u5217\u8868<\/h2>\n
    \n
  • \u4eba\u8138\u4fee\u590d<\/strong>\uff1a\u901a\u8fc7\u4ee3\u7801\u672c\u67e5\u627e\u53d8\u6362\u5668\u6280\u672f\uff0c\u63d0\u5347\u4f4e\u8d28\u91cf\u6216\u635f\u574f\u4eba\u8138\u56fe\u50cf\u7684\u6e05\u6670\u5ea6\u548c\u7ec6\u8282\u3002<\/li>\n
  • \u56fe\u50cf\u4e0a\u8272<\/strong>\uff1a\u4e3a\u9ed1\u767d\u6216\u892a\u8272\u7684\u56fe\u50cf\u6dfb\u52a0\u81ea\u7136\u7684\u989c\u8272\u3002<\/li>\n
  • \u56fe\u50cf\u4fee\u8865<\/strong>\uff1a\u4fee\u590d\u56fe\u50cf\u4e2d\u7684\u7f3a\u5931\u90e8\u5206\uff0c\u4f7f\u5176\u6062\u590d\u5b8c\u6574\u3002<\/li>\n
  • \u89c6\u9891\u5904\u7406<\/strong>\uff1a\u652f\u6301\u5bf9\u89c6\u9891\u4e2d\u7684\u4eba\u8138\u8fdb\u884c\u4fee\u590d\u548c\u589e\u5f3a\u3002<\/li>\n
  • \u5728\u7ebf\u6f14\u793a<\/strong>\uff1a\u63d0\u4f9b\u5728\u7ebf\u6f14\u793a\u529f\u80fd\uff0c\u7528\u6237\u53ef\u4ee5\u76f4\u63a5\u5728\u6d4f\u89c8\u5668\u4e2d\u4f53\u9a8c\u4fee\u590d\u6548\u679c\u3002<\/li>\n<\/ul>\n

     <\/p>\n

     <\/p>\n

    CodeFormer \u4f7f\u7528\u5e2e\u52a9<\/h2>\n

    \u5b89\u88c5\u6d41\u7a0b<\/h3>\n
      \n
    1. \u514b\u9686\u4ee3\u7801\u5e93<\/strong>\uff1a\n
      git clone<\/span> https:\/\/github.com\/sczhou\/CodeFormer\r\ncd<\/span> CodeFormer\r\n<\/code><\/pre>\n
       <\/li>\n
    2. \u521b\u5efa\u5e76\u6fc0\u6d3b\u865a\u62df\u73af\u5883<\/strong>\uff1a\n
      conda create -n codeformer python=3.8 -y\r\nconda activate codeformer\r\n<\/code><\/pre>\n
       <\/li>\n
    3. \u5b89\u88c5\u4f9d\u8d56<\/strong>\uff1a\n
      pip install -r requirements.txt\r\npython basicsr\/setup.py develop\r\nconda install -c conda-forge dlib\r\n<\/code><\/pre>\n
       <\/li>\n
    4. \u4e0b\u8f7d\u9884\u8bad\u7ec3\u6a21\u578b<\/strong>\uff1a\n
      python scripts\/download_pretrained_models.py facelib\r\npython scripts\/download_pretrained_models.py dlib\r\npython scripts\/download_pretrained_models.py CodeFormer\r\n<\/code><\/pre>\n
       <\/li>\n<\/ol>\n
      \u4f7f\u7528\u6d41\u7a0b<\/h3>\n
        \n
      1. \u51c6\u5907\u6d4b\u8bd5\u6570\u636e<\/strong>\uff1a \u5c06\u6d4b\u8bd5\u56fe\u50cf\u653e\u5165\u00a0inputs\/TestWhole<\/code>\u00a0\u6587\u4ef6\u5939\u4e2d\u3002\u5982\u679c\u9700\u8981\u6d4b\u8bd5\u88c1\u526a\u548c\u5bf9\u9f50\u7684\u4eba\u8138\u56fe\u50cf\uff0c\u53ef\u4ee5\u5c06\u5b83\u4eec\u653e\u5165\u00a0inputs\/cropped_faces<\/code>\u00a0\u6587\u4ef6\u5939\u4e2d\u3002<\/li>\n
      2. \u8fd0\u884c\u63a8\u7406\u4ee3\u7801<\/strong>\uff1a\n
        python inference_codeformer.py --input_path inputs\/TestWhole --output_path results\r\n<\/code><\/pre>\n
        \u8be5\u547d\u4ee4\u5c06\u5904\u7406\u00a0inputs\/TestWhole<\/code>\u00a0\u6587\u4ef6\u5939\u4e2d\u7684\u6240\u6709\u56fe\u50cf\uff0c\u5e76\u5c06\u7ed3\u679c\u4fdd\u5b58\u5728\u00a0results<\/code>\u00a0\u6587\u4ef6\u5939\u4e2d\u3002<\/li>\n<\/ol>\n
        \u529f\u80fd\u64cd\u4f5c\u6d41\u7a0b<\/h3>\n
          \n
        1. \u4eba\u8138\u4fee\u590d<\/strong>\uff1a\n
            \n
          • \u5c06\u5f85\u4fee\u590d\u7684\u56fe\u50cf\u653e\u5165\u00a0inputs\/TestWhole<\/code>\u00a0\u6587\u4ef6\u5939\u3002<\/li>\n
          • \u8fd0\u884c\u63a8\u7406\u4ee3\u7801\uff0c\u751f\u6210\u4fee\u590d\u540e\u7684\u56fe\u50cf\u3002<\/li>\n<\/ul>\n<\/li>\n
          • \u56fe\u50cf\u4e0a\u8272<\/strong>\uff1a\n
              \n
            • \u5c06\u9ed1\u767d\u56fe\u50cf\u653e\u5165\u00a0inputs\/TestWhole<\/code>\u00a0\u6587\u4ef6\u5939\u3002<\/li>\n
            • \u4f7f\u7528\u00a0inference_colorization.py<\/code>\u00a0\u811a\u672c\u8fdb\u884c\u4e0a\u8272\u5904\u7406\u3002<\/li>\n<\/ul>\n<\/li>\n
            • \u56fe\u50cf\u4fee\u8865<\/strong>\uff1a\n
                \n
              • \u5c06\u9700\u8981\u4fee\u8865\u7684\u56fe\u50cf\u653e\u5165\u00a0inputs\/TestWhole<\/code>\u00a0\u6587\u4ef6\u5939\u3002<\/li>\n
              • \u4f7f\u7528\u00a0inference_inpainting.py<\/code>\u00a0\u811a\u672c\u8fdb\u884c\u4fee\u8865\u5904\u7406\u3002<\/li>\n<\/ul>\n<\/li>\n
              • \u89c6\u9891\u5904\u7406<\/strong>\uff1a\n
                  \n
                • \u5c06\u89c6\u9891\u6587\u4ef6\u653e\u5165\u6307\u5b9a\u6587\u4ef6\u5939\u3002<\/li>\n
                • \u8fd0\u884c\u63a8\u7406\u4ee3\u7801\uff0c\u5904\u7406\u89c6\u9891\u4e2d\u7684\u4eba\u8138\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n
                   <\/p>\n
                  \u9762\u90e8\u4fee\u590d\uff08\u88c1\u526a\u548c\u5bf9\u9f50\u7684\u9762\u90e8\uff09<\/strong><\/p>\n
                  # For cropped and aligned faces
                  \npython inference_codeformer.py -w 0.5 –has_aligned –input_path [input folder]<\/p><\/blockquote>\n
                  \u6574\u4f53\u56fe\u50cf\u589e\u5f3a<\/strong><\/p>\n
                  # For whole image
                  \n# Add ‘–bg_upsampler realesrgan’ to enhance the background regions with Real-ESRGAN
                  \n# Add ‘–face_upsample’ to further upsample restorated face with Real-ESRGAN
                  \npython inference_codeformer.py -w 0.7 –input_path [image folder\/image path]<\/p><\/blockquote>\n
                  \u89c6\u9891\u589e\u5f3a<\/strong><\/p>\n
                  # For video clips
                  \npython inference_codeformer.py –bg_upsampler realesrgan –face_upsample -w 1.0 –input_path<\/p><\/blockquote>\n
                   <\/p>\n
                  \u4fdd\u771f\u5ea6\u6743\u91cdw\u4f4d\u4e8e [0, 1] \u4e2d\u3002\u901a\u5e38\uff0c\u8f83\u5c0f\u7684w\u5f80\u5f80\u4f1a\u4ea7\u751f\u8f83\u9ad8\u8d28\u91cf\u7684\u7ed3\u679c\uff0c\u800c\u8f83\u5927\u7684w\u4f1a\u4ea7\u751f\u8f83\u9ad8\u4fdd\u771f\u5ea6\u7684\u7ed3\u679c\u3002<\/p>\n
                  \u7ed3\u679c\u5c06\u4fdd\u5b58\u5728results\u6587\u4ef6\u5939\u4e2d\u3002<\/p>\n
                   <\/p>\n
                  CodeFormer \u4e00\u952e\u90e8\u7f72\u5305<\/h2>\n
                  \"AI\u751f\u4ea7\u529b\u5e94\u7528\"\u6b64\u5904\u5185\u5bb9\u5df2\u7ecf\u88ab\u4f5c\u8005\u9690\u85cf\uff0c\u8bf7\u8f93\u5165\u9a8c\u8bc1\u7801\u67e5\u770b\u5185\u5bb9<\/span>
                  \u9a8c\u8bc1\u7801\uff1a<\/span><\/form>
                  <\/div>\u8bf7\u5173\u6ce8\u672c\u7ad9\u5fae\u4fe1\u516c\u4f17\u53f7\uff0c\u56de\u590d\u201c\u9a8c\u8bc1\u7801<\/span>\u201d\uff0c\u83b7\u53d6\u9a8c\u8bc1\u7801\u3002\u5728\u5fae\u4fe1\u91cc\u641c\u7d22\u201cAI\u751f\u4ea7\u529b\u5e94\u7528<\/span>\u201d\u6216\u8005\u201cArtificial9527<\/span>\u201d\u6216\u8005\u5fae\u4fe1\u626b\u63cf\u53f3\u4fa7\u4e8c\u7ef4\u7801\u90fd\u53ef\u4ee5\u5173\u6ce8\u672c\u7ad9\u5fae\u4fe1\u516c\u4f17\u53f7\u3002<\/span>
                  <\/div><\/div><\/p>\n","protected":false},"excerpt":{"rendered":"
                  CodeFormer \u7efc\u5408\u4ecb\u7ecd CodeFormer \u662f\u4e00\u4e2a\u9762\u5411\u9c81\u68d2\u76f2\u8138\u4fee\u590d\u7684\u4ee3\u7801\u5e93\uff0c\u7531\u5357\u6d0b\u7406\u5de5\u5927\u5b66 S-Lab \u7684\u7814\u7a76\u56e2\u961f\u5f00\u53d1\uff0c\u5e76\u5728 NeurIPS 2022 \u4e0a\u53d1\u8868\u3002\u8be5\u9879\u76ee\u5229\u7528\u4ee3\u7801\u672c\u67e5\u627e\u53d8\u6362\u5668\uff08Codebook Lookup Trans…<\/p>\n","protected":false},"author":1,"featured_media":61023,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[410,440],"tags":[254,218,230],"class_list":["post-6569","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-art","category-image-restoration","tag-aifuyexiangmu","tag-aituxiangfangda","tag-aikaiyuanxiangmu"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/6569","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/comments?post=6569"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/6569\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media\/61023"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media?parent=6569"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/categories?post=6569"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/tags?post=6569"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}