{"id":2469,"date":"2022-12-01T00:04:55","date_gmt":"2022-11-30T16:04:55","guid":{"rendered":"http:\/\/divis.cn\/?p=2469"},"modified":"2026-02-07T19:13:08","modified_gmt":"2026-02-07T11:13:08","slug":"%e8%ae%a1%e7%ae%97%e6%9c%ba%e8%a7%86%e8%a7%89%e7%9a%84%e4%b8%89%e5%a4%a7%e5%9f%ba%e7%a1%80%e4%bb%bb%e5%8a%a1%e7%ae%80%e4%bb%8b","status":"publish","type":"post","link":"https:\/\/www.divis.cn\/index.php\/2022\/12\/01\/%e8%ae%a1%e7%ae%97%e6%9c%ba%e8%a7%86%e8%a7%89%e7%9a%84%e4%b8%89%e5%a4%a7%e5%9f%ba%e7%a1%80%e4%bb%bb%e5%8a%a1%e7%ae%80%e4%bb%8b\/","title":{"rendered":"\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u4e09\u5927\u57fa\u7840\u4efb\u52a1\u7b80\u4ecb"},"content":{"rendered":"

\u76ee\u524d\u672c\u4eba\u5728\u6d59\u5927\u8fdb\u4fee\uff0c\u611f\u53d7\u5230AI+VIS\u7684\u5e7f\u6cdb\u5e94\u7528\u524d\u666f\u3002\u4e4b\u524d\u5df2\u7ecf\u5199\u4e86\u535a\u5ba2\u7b80\u5355\u4ecb\u7ecd\u8fc7\u81ea\u7136\u8bed\u8a00\u5904\u7406\uff08NLP\uff09\u7684\u5185\u5bb9\u3002\u8fd9\u6b21\u5219\u4ece\u5bf9\u56fe\u50cf\u7684\u7406\u89e3\u51fa\u53d1\uff0c\u7b80\u5355\u4ecb\u7ecd\u8ba1\u7b97\u673a\u89c6\u89c9\u3002<\/p>\n

\n

\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u6982\u5ff5\uff0c\u4e0e\u673a\u5668\u89c6\u89c9\u7684\u533a\u522b<\/h3>\n

\u8ba1\u7b97\u673a\u89c6\u89c9\uff08computer vision\uff09\u65e8\u5728\u8bc6\u522b\u548c\u7406\u89e3\u56fe\u50cf\/\u89c6\u9891\u4e2d\u7684\u5185\u5bb9\u3002
\n\u4e0e\u4e4b\u7c7b\u4f3c\u7684\u8bcd\u6c47\u662f\u673a\u5668\u89c6\u89c9\uff08machine vision\uff09\uff0c \u4f46\u662f\u673a\u5668\u89c6\u89c9\u66f4\u504f\u91cd\u4e8e\u8ba1\u7b97\u673a\u89c6\u89c9\u6280\u672f\u7684\u5de5\u7a0b\u5e94\u7528\uff0c\u800c\u8ba1\u7b97\u673a\u89c6\u89c9\u5219\u66f4\u504f\u5411\u5b66\u672f\u7684\u7b97\u6cd5\u7814\u7a76\u3002\u7b80\u5355\u7684\u8bf4\uff0c\u8ba1\u7b97\u673a\u89c6\u89c9\u76f8\u5f53\u4e8e\u8bbe\u8ba1API\uff0c\u673a\u5668\u89c6\u89c9\u76f8\u5f53\u4e8e\u8c03\u7528API\u5f00\u53d1\u8f6f\u4ef6\u3002\u5982\u679c\u53ea\u662f\u5c31\u56fe\u50cf\u5185\u5bb9\u5e94\u5982\u4f55\u7406\u89e3\u7684\u4e00\u822c\u6027\u8ba8\u8bba\uff0c\u5219\u4e24\u4e2a\u8bcd\u53ef\u4ee5\u6df7\u7528\u3002<\/p>\n

\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u9876\u7ea7\u4f1a\u8bae\u6709CVPR\u3001ICCV\u3001\u548cECCV\uff0c\u6b64\u5916ICLR\u4e5f\u6709\u4e0d\u5c11\u8ba1\u7b97\u673a\u89c6\u89c9\u8bba\u6587\u3002\u9876\u7ea7\u671f\u520a\u6709IJCV\u548cTPAMI\u3002CVPR\u5e74\u6295\u7a3f\u91cf\u4ece2020\u5e74\u5c31\u8d85\u8fc71\u4e07\u7bc7\uff0c\u5f55\u7528\u91cf\u57282022\u5e74\u4e5f\u8d85\u8fc7\u4e862000\uff0c\u53ef\u89c1\u706b\u70ed\u7a0b\u5ea6\u3002\u7531\u4e8e\u8ba1\u7b97\u673a\u89c6\u89c9\u9886\u57df\u53d1\u5c55\u5341\u5206\u8fc5\u901f\uff0c\u4e0d\u8bba\u8eab\u5904\u5b66\u672f\u754c\u6216\u4ea7\u4e1a\u754c\uff0c\u901a\u8fc7\u9605\u8bfb\u9876\u7ea7\u4f1a\u8bae\u548c\u671f\u520a\u8bba\u6587\u4e86\u89e3\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u6700\u8fd1\u7814\u7a76\u6210\u679c\u90fd\u5341\u5206\u5fc5\u8981\u3002<\/p>\n

\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u57fa\u7840\u4efb\u52a1<\/h3>\n

\u867d\u7136\u8bf4\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u76ee\u7684\u662f\u7406\u89e3\u56fe\u50cf\/\u89c6\u9891\u4e2d\u7684\u5185\u5bb9\uff0c\u4f46\u600e\u4e48\u7b97\u662f\u7406\u89e3\uff0c\u9700\u8981\u6709\u5177\u4f53\u7684\u4efb\u52a1\u8fdb\u884c\u63cf\u8ff0\u3002\u8ba1\u7b97\u673a\u79d1\u5b66\u5bb6\u603b\u7ed3\u4e86\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u57fa\u7840\u4efb\u52a1\uff0c\u5305\u542b\u5982\u4e0b\u4e09\u5927\u7c7b\uff0c\u5206\u522b\u662f\uff1a<\/p>\n

    \n
  1. \u56fe\u50cf\u5206\u7c7b\uff08Image Classification\uff09<\/li>\n
  2. \u76ee\u6807\u8bc6\u522b\uff1a\u76ee\u6807\u5b9a\u4f4d\uff08Object Localization\uff09– \u76ee\u6807\u68c0\u6d4b\uff08Object Detection\uff09<\/li>\n
  3. \u56fe\u50cf\u5206\u5272\uff1a\u8bed\u4e49\u5206\u5272\uff08Semantic Segmentation\uff09– \u5b9e\u4f8b\u5206\u5272\uff08Instance Segmentation\uff09–\u5168\u666f\u5206\u5272\uff08Panoptic Segmentation\uff09<\/li>\n<\/ol>\n

    \u4ee5\u4e0a\u4e09\u7c7b\u57fa\u7840\u4efb\u52a1\u88ab\u7edf\u4e00\u5f52\u5165\u4e86\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u201c\u56fe\u50cf\u7406\u89e3<\/strong>\u201d\u8303\u7574\u3002\u8fd9\u662f\u56e0\u4e3a\u73b0\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u8fd8\u6709\u5f88\u591a\u5176\u4ed6\u5e94\u7528\u65b9\u5411\uff0c\u6bd4\u5982\u56fe\u50cf\u751f\u6210<\/strong>\uff08\u8d85\u5206\u8fa8\u7387\u3001\u6587\u672cto\u56fe\u50cf\u3001\u56fe\u50cf\u53bb\u566a\u3001\u98ce\u683c\u8fc1\u79fb\uff0c\u89c6\u9891\u751f\u6210\uff09\uff0c\u56fe\u50cf\u68c0\u7d22<\/strong>\uff08\u4ee5\u56fe\u641c\u56fe\u3001\u4ee5\u6587\u641c\u56fe\uff0c\u89c6\u9891\u641c\u7d22\uff09\u56fe\u50cf\u5e8f\u5217<\/b>\uff08\u76ee\u6807\u8ddf\u8e2a\u3001\u56fe\u50cf\u5e8f\u5217\u5206\u7c7b\u3001\u6b65\u6001\u8bc6\u522b\uff09\uff0c\u591a\u6a21\u6001<\/b>\uff08\u4e0eNLP\u7ed3\u5408\uff0c\u4e0e\u4e13\u4e1a\u9886\u57df\u77e5\u8bc6\u7ed3\u5408\u7b49\uff09\u3002\u4f46\u662f\u540e\u9762\u6240\u8bf4\u8fd9\u4e9b\u5e94\u7528\u65b9\u5411\uff0c\u65e0\u4e00\u4e0d\u662f\u4ee5\u56fe\u50cf\u7406\u89e3\u4e3a\u57fa\u7840\u7684\u3002\u56e0\u6b64\u5fc5\u987b\u5bf9\u56fe\u50cf\u7406\u89e3\u7684\u4e09\u5927\u4efb\u52a1\u6709\u6240\u4e86\u89e3\u3002<\/p>\n

    <\/p>\n

    \u56fe\u50cf\u5206\u7c7b<\/h3>\n

    \u56fe\u50cf\u5206\u7c7b\u4e3b\u8981\u89e3\u51b3\u56fe\u7247\u201c\u662f\u4ec0\u4e48\uff1f\u201d\u7684\u95ee\u9898\uff0c\u5373\u7ed9\u5b9a\u4e00\u5f20\u56fe\u6216\u4e00\u6bb5\u89c6\u9891\uff0c\u5224\u65ad\u91cc\u9762\u5305\u542b\u4ec0\u4e48\u7c7b\u522b\u7684\u76ee\u6807\u3002\u56fe\u50cf\u5206\u7c7b\u4f5c\u4e3a\u673a\u5668\u89c6\u89c9\u6700\u57fa\u7840\u7684\u95ee\u9898\uff0c\u662f\u65e9\u671f\u4eba\u5de5\u667a\u80fd\u4e3b\u8981\u7684\u6311\u6218\u4efb\u52a1\u3002\u56fe\u50cf\u5206\u7c7b\u7684\u5e38\u7528\u6570\u636e\u96c6\u662fImageNet\uff0c\u5305\u542b1400\u4e07\u5f20\u56fe\u50cf\u548c1\u4e07\u591a\u4e2a\u7c7b\u522b\u30022012\u5e74\uff0c\u91c7\u7528CNN\u6784\u5efa\u7684\u6df1\u5ea6\u5b66\u4e60\u7f51\u7edcALexNet\u83b7\u5f97\u4e86ImageNet\u56fe\u50cf\u8bc6\u522b\u6bd4\u8d5b\u51a0\u519b\uff0c\u88ab\u8ba4\u4e3a\u662f\u6df1\u5ea6\u5b66\u4e60\u63a8\u5e7f\u5143\u5e74\u3002\u73b0\u4eca\u9664\u4e86CNN\uff0c\u6765\u81eaNLP\u7684Transformer\u67b6\u6784\u7684\u6df1\u5ea6\u5b66\u4e60\u7f51\u7edc\u4f53\u73b0\u4e86\u66f4\u5f3a\u6027\u80fd\u3002<\/p>\n

    \u56fe\u50cf\u5206\u7c7b\u4f5c\u4e3a\u6574\u4e2a\u8ba1\u7b97\u673a\u89c6\u89c9\u5927\u53a6\u7684\u57fa\u7840\uff0c\u957f\u671f\u5904\u4e8e\u91cd\u8981\u7684\u4f4d\u7f6e\u3002\u81ea2017\u5e74ImageNet\u4e2d\u5927\u90e8\u5206\u673a\u5668\u5b66\u4e60\u6a21\u578b\u90fd\u53d6\u5f97\u4e8695%\u4ee5\u4e0a\u7684\u51c6\u786e\u7387\u540e\uff0c\u9488\u5bf9\u666e\u901a\u56fe\u50cf\u7684\u56fe\u50cf\u5206\u7c7b\u5df2\u4e0d\u5728\u662f\u91cd\u70b9\u3002\u8fd1\u5e74\u6765\u4e3b\u8981\u7814\u7a76\u96c6\u4e2d\u4e8e\u5c0f\u6837\u672c\u6570\u636e\u548c\u98ce\u683c\u8fc1\u79fb\u540e\u7684\u56fe\u50cf\u5206\u7c7b\u7b49\u7ec6\u5206\u9886\u57df\u3002<\/p>\n

    \u76ee\u6807\u8bc6\u522b\uff1a\u76ee\u6807\u5b9a\u4f4d\uff08Object Localization\uff09– \u76ee\u6807\u68c0\u6d4b\uff08Object Detection\uff09<\/h3>\n

    \u5728\u56fe\u50cf\u5206\u7c7b\u7684\u57fa\u7840\u4e0a\uff0c\u6211\u4eec\u8fd8\u60f3\u77e5\u9053\u76ee\u6807\uff08\u6bd4\u5982\u82f9\u679c\uff09\u5177\u4f53\u5728\u53e6\u4e00\u5f20\u56fe\uff08\u6bd4\u5982\u4e00\u5f20\u9759\u7269\u753b\uff09\u7684\u4ec0\u4e48\u4f4d\u7f6e\u3002\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u4e2d\uff0c\u76ee\u6807\u7269\u4f53\u7684\u4f4d\u7f6e\u901a\u5e38\u662f\u4ee5\u5305\u56f4\u76d2\u7684(bounding box)\u5f62\u5f0f\u6807\u51fa\u3002\u4ece\u63cf\u8ff0\u4e2d\u6211\u4eec\u53ef\u4ee5\u53d1\u73b0\uff0c\u8981\u5b8c\u6210\u76ee\u6807\u5b9a\u4f4d\u81f3\u5c11\u8981\u5b8c\u6210\u4e24\u4e2a\u5b50\u4efb\u52a1\uff0c\u4e00\u4e2a\u5b50\u4efb\u52a1\u662f\u5224\u65ad\u4ec0\u4e48\u662f\u76ee\u6807\u7269\u4f53\uff0c\u901a\u5e38\u8981\u7528\u56fe\u50cf\u5206\u7c7b\u65b9\u6cd5\u5c06\u56fe\u7247\u5206\u4e3a\u524d\u666f\uff08\u76ee\u6807\u7269\u4f53\uff09\u548c\u80cc\u666f\uff1b\u53e6\u4e00\u4e2a\u5b50\u4efb\u52a1\u662f\u5224\u65ad\u76ee\u6807\u7269\u4f53\u4f4d\u7f6e\u3002<\/p>\n

    \u968f\u7740\u8ba1\u7b97\u673a\u662f\u89c9\u7684\u4e0d\u65ad\u53d1\u5c55\uff0c\u76ee\u6807\u5b9a\u4f4d\uff08Object Localization\uff09<\/strong>\u8fd9\u4e00\u540d\u8bcd\u9010\u6e10\u4e3a\u76ee\u6807\u68c0\u6d4b\uff08Object Detection\uff09<\/strong>\u6240\u53d6\u4ee3\u3002\u5728\u76ee\u6807\u5b9a\u4f4d\u4e2d\uff0c\u901a\u5e38\u53ea\u6709\u4e00\u4e2a\u6216\u56fa\u5b9a\u6570\u76ee\u7684\u76ee\u6807\uff0c\u800c\u76ee\u6807\u68c0\u6d4b\u66f4\u4e00\u822c\u5316\uff0c\u5176\u56fe\u50cf\u4e2d\u51fa\u73b0\u7684\u76ee\u6807\u79cd\u7c7b\u548c\u6570\u76ee\u90fd\u4e0d\u5b9a\u3002\u4e5f\u5c31\u662f\u8bf4\uff0c\u5b9a\u4f4d\u4e0e\u68c0\u6d4b\u7684\u4e3b\u8981\u533a\u522b\u5728\u4e8e\u5206\u7c7b\u8fd9\u4e00\u5b50\u4efb\u52a1\u7684\u590d\u6742\u6027\u3002\u5982\u679c\u53ea\u9700\u8981\u5224\u65ad\u4e00\u79cd\u76ee\u6807\u7269\u4f53\u7684\u4f4d\u7f6e\uff0c\u5373\u5176\u5206\u7c7b\u5b50\u4efb\u52a1\u662f\u4e00\u4e2a\u4e8c\u5206\u7c7b\u4efb\u52a1\uff08\u53ea\u9700\u8981\u5224\u65ad\u524d\u666f\u548c\u80cc\u666f\uff09\uff0c\u90a3\u5c31\u662f\u76ee\u6807\u5b9a\u4f4d\u3002\u5982\u679c\u8fd8\u9700\u8981\u8fdb\u4e00\u6b65\u77e5\u9053\u76ee\u6807\u7269\u4f53\u662f\u4ec0\u4e48\uff08\u662f\u4eba\uff1f\u662f\u8f66\uff1f\u662f\u732b\uff1f\u662f\u72d7\uff1f\u7b49\u7b49\uff09\uff0c\u5373\u5176\u5206\u7c7b\u5b50\u4efb\u52a1\u662f\u4e00\u4e2a\u591a\u5206\u7c7b\u4efb\u52a1\uff0c\u90a3\u5c31\u662f\u76ee\u6807\u68c0\u6d4b\u3002<\/p>\n

    <\/p>\n

    \u663e\u7136\uff0c\u76ee\u6807\u68c0\u6d4b\u662f\u6bd4\u76ee\u6807\u5b9a\u4f4d\u66f4\u5177\u6311\u6218\u6027\u7684\u4efb\u52a1\uff0c\u4e5f\u662f\u73b0\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u9886\u57df\u7684\u7814\u7a76\u70ed\u70b9\u4e4b\u4e00\u3002\u73b0\u4eca\u5e38\u7528\u7684\u6570\u636e\u96c6\u662f\u5fae\u8f6f\u7684COCO\uff0c\u5176\u5305\u542b80k\u8bad\u7ec3\u56fe\u50cf\u300140k\u9a8c\u8bc1\u56fe\u50cf\u3001\u548c20k\u6ca1\u6709\u516c\u5f00\u6807\u8bb0\u7684\u6d4b\u8bd5\u56fe\u50cf(test-dev)\uff0c80\u4e2a\u7c7b\u522b\uff0c\u5e73\u5747\u6bcf\u5f20\u56fe7.2\u4e2a\u76ee\u6807\u3002<\/p>\n

    \u5173\u4e8e\u76ee\u6807\u68c0\u6d4b\u5177\u4f53\u7684\u65b9\u6cd5\uff0c\u5219\u6709\u4e24\u79cd\u601d\u8def\uff1a\u4e00\u662f\u4e24\u7ea7\u5f0f\u6846\u67b6\uff0c\u5148\u5224\u65ad\u56fe\u50cf\u7684\u989c\u8272\u3001\u7eb9\u7406\u3001\u9762\u79ef\u3001\u4f4d\u7f6e\u7b49\uff0c\u5408\u5e76\u76f8\u4f3c\u7684\u50cf\u7d20\uff0c\u5f97\u5230\u4e00\u4e9b\u5019\u9009\u533a\u57df\uff0c\u7136\u540e\u7528\u6ed1\u52a8\u7a97\u53e3\u4f9d\u6b21\u628a\u6240\u6709\u5019\u9009\u533a\u57df\u5339\u914d\u4e00\u904d\u3002\u5178\u578b\u5982Fast R-CNN\u3002\u4e8c\u662f\u5355\u7ea7\u5f0f\u6846\u67b6\uff0c\u5c31\u662f\u4e0d\u9884\u5148\u751f\u6210\u5019\u9009\u533a\u57df\uff0c\u800c\u662f\u5c06\u56fe\u50cf\u7b80\u5355\u5212\u5206\u4e3a\u7f51\u683c\uff0c\u7136\u540e\u76f4\u63a5\u8f93\u51fa\u5206\u7c7b\/\u56de\u5f52\u7ed3\u679c\uff0c\u5178\u578b\u5982YOLO1\uff0c2\uff0c3\uff0c4\uff0c5\u3002\u4f46\u65e0\u8bba\u4f55\u79cd\u601d\u8def\uff0c\u8fd8\u6ca1\u6709\u54ea\u4e2a\u76ee\u6807\u68c0\u6d4b\u6a21\u578b\u80fd\u5728COCO\u4e0a\u8fbe\u523095%\u8fd9\u6837\u51c6\u786e\u7387\u3002<\/p>\n

    \u76ee\u6807\u68c0\u6d4b\u5728\u7279\u5b9a\u9886\u57df\u4e2d\u6709\u4e00\u4e9b\u91cd\u8981\u5e94\u7528\uff0c\u4f8b\u5982\u4eba\u8138\u68c0\u6d4b\u3001\u884c\u4eba\u68c0\u6d4b\u3001\u6587\u672c\u68c0\u6d4b\u3001\u4ea4\u901a\u4fe1\u53f7\u548c\u9065\u611f\u76ee\u6807\u68c0\u6d4b\u7b49\u3002\u8fd9\u90fd\u9700\u8981\u9488\u5bf9\u7279\u5b9a\u95ee\u9898\u8fdb\u884c\u7279\u6b8a\u5904\u7406\u548c\u4f18\u5316\u3002\u9664\u6b64\u4ee5\u5916\uff0c\u878d\u5408\u591a\u79cd\u6570\u636e\u6e90\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\u4e5f\u662f\u672a\u6765\u7684\u53d1\u5c55\u8d8b\u52bf\u3002\u4f8b\u5982\u878d\u5408\u6587\u672c\u4fe1\u606f\u3001\u6fc0\u5149\u96f7\u8fbe\u548c\u56fe\u50cf\u7b49\u591a\u6a21\u6001\u6570\u636e\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\uff0c\u5bf9\u4e8e\u81ea\u52a8\u9a7e\u9a76\u3001\u65e0\u4eba\u673a\u3001\u56fe\u50cf\u6570\u636e\u67e5\u8be2\u7b49\u975e\u5e38\u91cd\u8981\u3002\u5173\u4e8e\u76ee\u6807\u68c0\u6d4b\u7684\u66f4\u591a\u4ecb\u7ecd\u53ef\u4ee5\u53c2\u8003\u8fd9\u7bc7\u7efc\u8ff0A Survey of Modern Deep Learning based Object Detection Models<\/a>\u3002<\/p>\n

    \u56fe\u50cf\u5206\u5272\uff1a\u8bed\u4e49\u5206\u5272\uff08Semantic Segmentation\uff09– \u5b9e\u4f8b\u5206\u5272\uff08Instance Segmentation\uff09–\u5168\u666f\u5206\u5272\uff08Panoptic Segmentation\uff09<\/h3>\n

    \u56fe\u50cf\u5206\u5272\u662f\u76ee\u6807\u68c0\u6d4b\u66f4\u8fdb\u9636\u7684\u4efb\u52a1\uff0c\u5b83\u9700\u8981\u68c0\u6d4b\u5230\u76ee\u6807\u7269\u4f53\uff0c\u7136\u540e\u628a\u76ee\u6807\u7269\u4f53\u4ece\u539f\u59cb\u56fe\u50cf\u4e2d\u5206\u51fa\u6765\u3002\u5982\u540c\u4ece\u76ee\u6807\u5b9a\u4f4d\u5230\u76ee\u6807\u68c0\u6d4b\u7684\u5347\u7ea7\u8fc7\u7a0b\uff0c\u56fe\u50cf\u5206\u5272\u7684\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e5f\u7ecf\u5386\u4e86\u4ece\u8bed\u4e49\u5206\u5272\u5230\u5b9e\u4f8b\u5206\u5272\u7684\u8fed\u4ee3\u3002<\/p>\n

    <\/p>\n

    \u5148\u8bf4\u8bed\u4e49\u5206\u5272\uff0c\u8bed\u4e49\u56fe\u50cf\u5206\u5272\u7684\u76ee\u6807\u662f\u7ed9\u56fe\u50cf\u4e2d\u7684\u6bcf\u4e2a\u50cf\u7d20\u90fd\u8d34\u4e0a\u5bf9\u5e94\u7684\u4e00\u7c7b\u6240\u4ee3\u8868\u7684\u5185\u5bb9\u7684\u6807\u7b7e\uff08stuff\uff09\u3002\u5176\u8f93\u51fa\u7684\u6700\u7ec8\u7ed3\u679c\u6709\u4e24\u79cd\u5448\u73b0\u65b9\u5f0f\uff1a\u533a\u57df\u63a9\u7801\u548c\u8f6e\u5ed3\u7ebf\u3002\u533a\u57df\u63a9\u7801\u5c31\u662f\u76f4\u63a5\u5c06\u5206\u597d\u7c7b\u522b\u7684\u50cf\u7d20\u7528\u533a\u57df\u63a9\u7801\u8868\u793a\uff0c\u5982\u679c\u5bf9\u533a\u57df\u63a9\u7801\u6307\u5b9a\u989c\u8272\uff0c\u90a3\u4e48\u8f93\u51fa\u7684\u7ed3\u679c\u5c06\u53ef\u4ee5\u770b\u5230\u4e00\u7247\u4e00\u7247\u7684\u8272\u5757\uff1b\u8f6e\u5ed3\u7ebf\u662f\u5728\u533a\u57df\u63a9\u7801\u7684\u57fa\u7840\u4e0a\uff0c\u5c06\u533a\u57df\u8fb9\u7f18\u7528\u4e00\u7cfb\u5217\u70b9\u8868\u793a\u6210\u4e00\u4e2a\u95ed\u5408\u591a\u8fb9\u5f62\u3002\u4ece\u8fd9\u4e2a\u89d2\u5ea6\u4e0a\u770b\uff0c\u8bed\u4e49\u5206\u5272\u80fd\u591f\u4f7f\u7528\u7ec6\u81f4\u8f6e\u5ed3\u7ebf\u5c06\u76ee\u6807\u7269\u4f53\u4ece\u539f\u59cb\u56fe\u50cf\u4e2d\u5708\u51fa\u6765\uff0c\u800c\u4e0d\u662f\u50cf\u76ee\u6807\u68c0\u6d4b\u90a3\u6837\u7b80\u5355\u5730\u7528\u65b9\u5f62\u8fb9\u754c\u6846\u6807\u8bb0\u3002\u5fc5\u987b\u6307\u51fa\u7684\u662f\uff0c\u8bed\u4e49\u5206\u5272\u867d\u7136\u533a\u5206\u4e86\u8272\u5757\uff0c\u4f46\u5e76\u4e00\u5b9a\u4e0d\u77e5\u9053\u8fd9\u4e2a\u8272\u5757\u53ef\u80fd\u4ee3\u8868\u4ec0\u4e48\u7269\u4f53\uff08thing\uff09\uff0c\u8fd9\u662f\u4e0e\u76ee\u6807\u68c0\u6d4b\u5927\u4e0d\u76f8\u540c\u7684\u5730\u65b9\u3002<\/p>\n

    \"\"
    \u57fa\u4e8e\u8bed\u4e49\u5206\u5272\u7684\u7b80\u6d01\u7ebf\u6761\u8096\u50cf\u753b\u751f\u6210\u65b9\u6cd5<\/a>\uff1a\u4f7f\u7528\u8bed\u4e49\u5206\u5272\u65b9\u6cd5\u5148\u83b7\u53d6\u533a\u57df\u63a9\u7801\uff0c\u518d\u63d0\u53d6\u8f6e\u5ed3\u7ebf\u3002\u8fc7\u7a0b\u4e2d\u63a7\u5236\u7ec6\u8282\u8868\u73b0\u7a0b\u5ea6\u5f97\u5230\u4e0d\u540c\u7cbe\u7ec6\u5ea6\u7684\u7ebf\u6761\u3002<\/figcaption><\/figure>\n

    \u8bed\u4e49\u5206\u5272\u7684\u5178\u578b\u5e94\u7528\u5982\u4ece\u4eba\u8138\u8bc6\u522b\u3001\u624b\u5199\u5b57\u7b26\u8bc6\u522b\u3001\u667a\u80fd\u62a0\u56fe\u3001\u66ff\u6362\u89c6\u9891\u4f1a\u8bae\u65f6\u7684\u80cc\u666f\u7b49\u201c\u5c0f\u5de5\u5177\u201d\uff0c\u5230\u6c7d\u8f66\u81ea\u52a8\u9a7e\u9a76\uff0c\u533b\u5b66\u5f71\u50cf\u81ea\u52a8\u8bc6\u522b\u7b49\u4e13\u4e1a\u5e94\u7528\u4e0d\u4e00\u800c\u8db3\u3002\u8bed\u4e49\u5206\u5272\u7684\u5177\u4f53\u65b9\u6cd5\uff0c\u4ece\u7b80\u5355\u7684\u8bbe\u7f6e\u9608\u503c\uff0c\u8fb9\u7f18\u68c0\u6d4b\uff0c\u805a\u7c7b\uff0c\u533a\u57df\u751f\u957f\u518d\u5230\u590d\u6742\u4e00\u70b9\u7684\u65b9\u6cd5\u8bf8\u5982\u4e3b\u52a8\u8f6e\u5ed3\u7ebf\u3002\u4ee3\u8868\u6027\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u662fFCN\uff082014\uff09\uff0cDeepLab\u7cfb\u5217\u7b49\u3002<\/p>\n

    \u8bed\u4e49\u5206\u5272\u662f\u662f\u4ece\u50cf\u7d20\u7ea7\u522b\u6765\u7406\u89e3\u56fe\u50cf\u7684\uff0c\u5b83\u53ea\u80fd\u533a\u5206\u8bed\u4e49\u5c42\u9762\u7684\u524d\u666f\u548c\u80cc\u666f\u5185\u5bb9\uff08stuff\uff09\uff0c\u4e0d\u80fd\u533a\u5206\u5c5e\u4e8e\u76f8\u540c\u7c7b\u522b\u7684\u4e0d\u540c\u5b9e\u4f8b\u3002\u4f8b\u5982\uff0c\u5f53\u56fe\u50cf\u4e2d\u6709\u591a\u53ea\u7f8a\u65f6\uff0c\u8bed\u4e49\u5206\u5272\u4f1a\u4e09\u53ea\u7f8a\u6574\u4f53\u7684\u6240\u6709\u50cf\u7d20\u9884\u6d4b\u4e3a\u201c\u7f8a\u201d\u8fd9\u4e2a\u7c7b\u522b\uff0c\u5c06\u5176\u4e0e\u80cc\u666f\u533a\u5206\u5f00\u3002\u4f46\u5b83\u4e0d\u80fd\u533a\u5206\u51fa\u54ea\u4e9b\u50cf\u7d20\u5c5e\u4e8e\u7b2c\u4e00\u53ea\u7f8a\u3001\u54ea\u4e9b\u50cf\u7d20\u5c5e\u4e8e\u7b2c\u4e8c\u53ea\u7f8a\u3002\u4e3a\u4e86\u4ee4\u8ba1\u7b97\u673a\u89c6\u89c9\u66f4\u52a0\u63a5\u8fd1\u4eba\u7684\u7406\u89e3\uff0c\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u7814\u7a76\u4eba\u5458\u63d0\u51fa\u4e86\u5b9e\u4f8b\u5206\u5272<\/strong>\u7684\u6982\u5ff5\uff0c\u8981\u6c42\u4e0d\u4ec5\u80fd\u533a\u5206\u7f8a\u6240\u5360\u7684\u50cf\u7d20\uff0c\u8fd8\u80fd\u591f\u533a\u5206\u76ee\u6807\u7269\u4f53\uff08thing\uff09\u3002<\/p>\n

    <\/p>\n

    \u56e0\u4e3a\u7ed3\u5408\u4e86\u8bed\u4e49\u5206\u5272\u548c\u76ee\u6807\u68c0\u6d4b\u4e24\u65b9\u9762\u7684\u529f\u80fd\uff0c\u5b9e\u4f8b\u5206\u5272\u7684\u5b9e\u73b0\u5219\u66f4\u590d\u6742\u4e00\u4e9b\u3002\u5177\u4f53\u5b9e\u73b0\u601d\u8def\u53ef\u4ee5\u5206\u4e3a\u4e24\u6761\uff1a\u5148\u76ee\u6807\u68c0\u6d4b\u518d\u8bed\u4e49\u5206\u5272\uff08\u4e5f\u88ab\u79f0\u4e3a\u81ea\u4e0b\u800c\u4e0a\u65b9\u6cd5\uff09\uff0c\u548c\u5148\u8bed\u4e49\u5206\u5272\u518d\u76ee\u6807\u68c0\u6d4b(\u4e5f\u88ab\u79f0\u4e3a\u81ea\u4e0b\u800c\u4e0a\u65b9\u6cd5)\u3002\u524d\u8005\u5148\u7528\u76ee\u6807\u68c0\u6d4b\u65b9\u6cd5\u5c06\u56fe\u50cf\u4e2d\u7684\u4e0d\u540c\u5b9e\u4f8b\u6846\u51fa\uff0c\u518d\u7528\u8bed\u4e49\u5206\u5272\u65b9\u6cd5\u5728\u4e0d\u540c\u5305\u56f4\u76d2\u5185\u8fdb\u884c\u9010\u50cf\u7d20\u6807\u8bb0\uff0c\u5178\u578b\u5982DeepMask\uff0cDeepSnake\u7b49 \u3002\u540e\u8005\u9996\u5148\u8fdb\u884c\u50cf\u7d20\u7ea7\u522b\u7684\u8bed\u4e49\u5206\u5272\uff0c\u518d\u901a\u8fc7\u805a\u7c7b\u3001\u5ea6\u91cf\u5b66\u4e60\u7b49\u624b\u6bb5\u533a\u5206\u4e0d\u540c\u7684\u5b9e\u4f8b\uff0c\u5178\u578b\u5982Mask R-CNN\uff0c Polar Mask\u3002<\/p>\n

    <\/p>\n

    \u6700\u540e\u8fd8\u6709\u4e00\u4e2a2018\u5e74\u4ee5\u540e\u63d0\u51fa\u7684\u65b0\u6982\u5ff5\uff0c\u5168\u666f\u5206\u5272<\/strong>\uff08Panoptic Segmentation\uff09\u3002\u5168\u666f\u5206\u5272\u5c06\u8bed\u4e49\u5206\u5272\uff08\u4e3a\u6bcf\u4e2a\u50cf\u7d20\u5206\u914d\u4e00\u4e2a\u7c7b\u6807\u7b7e\uff09\u548c\u5b9e\u4f8b\u5206\u5272\uff08\u68c0\u6d4b\u5e76\u5206\u5272\u6bcf\u4e2a\u5bf9\u8c61\u5b9e\u4f8b\uff09\u7684\u5178\u578b\u4efb\u52a1\u7edf\u4e00\u8d77\u6765\uff0c\u65e8\u5728\u540c\u65f6\u5206\u5272\u5b9e\u4f8b\u5c42\u9762\u7684\u76ee\u6807\uff08thing\uff09\u548c\u8bed\u4e49\u5c42\u9762\u7684\u80cc\u666f\u5185\u5bb9\uff08stuff\uff09\uff0c\u5c06\u8f93\u5165\u56fe\u50cf\u4e2d\u7684\u6bcf\u4e2a\u50cf\u7d20\u70b9\u90fd\u8d4b\u4e88\u7c7b\u522b\u6807\u7b7e\u548c\u5b9e\u4f8bID\uff0c\u751f\u6210\u5168\u5c40\u7684\uff0c\u7edf\u4e00\u7684\u5206\u5272\u56fe\u50cf\u3002\u5168\u666f\u5206\u5272\u5728\u6a21\u62df\u4eba\u7c7b\u89c6\u89c9\u7cfb\u7edf\u7684\u9053\u8def\u4e0a\u53ef\u8c13\u53c8\u8fdb\u4e00\u6b65\u3002<\/p>\n

    <\/p>\n

    <\/p>\n

    \u5168\u666f\u5206\u5272\u5728\u5177\u4f53\u5b9e\u73b0\u4e0a\u6709\u4e09\u79cd\u65b9\u6cd5\u30021.thing\u548cstuff\u4f7f\u7528\u4e0d\u540c\u7684\u5206\u652f\u9884\u6d4b\uff0c\u5176\u4e2dthing\u7684\u5206\u5272\u57fa\u4e8e\u76ee\u6807\u68c0\u6d4b\u8fb9\u754c\u6846\uff0c\u5982Panoptic FPN\u30022.thing\u548cstuff\u4f7f\u7528\u4e0d\u540c\u7684\u5206\u652f\u9884\u6d4b\uff0c\u5148\u9884\u6d4b\u8bed\u4e49\u518d\u751f\u6210\u5b9e\u4f8b\uff0c\u4e0d\u9700\u8981\u5148\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\uff0c\u66f4\u4e0d\u9700\u8981\u4f9d\u8d56\u8fb9\u754c\u6846\uff0c\u5982Panoptic-DeepLab\u30023.thing\u548cstuff\u5b8c\u5168\u4f7f\u7528\u76f8\u540c\u7684\u7ed3\u6784\u8fdb\u884c\u9884\u6d4b\uff0c\u5982Panoptic-FCN\u3002<\/p>\n

    \u56e0\u4e3a\u662f\u76ee\u6807\u68c0\u6d4b\u7684\u8fdb\u9636\u4efb\u52a1\uff0c\u6240\u4ee5\u8bed\u4e49\u5206\u5272\u3001\u5b9e\u4f8b\u5206\u5272\u3001\u5168\u666f\u5206\u5272\u4e5f\u901a\u5e38\u4f7f\u7528COCO\u6570\u636e\u96c6\u8fdb\u884c\u8bad\u7ec3\u3002\u5f53\u7136\u8fd8\u6709\u522b\u7684\u6570\u636e\u96c6\uff0c\u5177\u4f53\u53ef\u4ee5\u53c2\u9605\u8fd9\u7bc7\u6587\u7ae0<\/a>\u3002\u5173\u4e8e\u56fe\u50cf\u5206\u5272\u7684\u673a\u5668\u5b66\u4e60\u6a21\u578b\u7684\u66f4\u591a\u4ecb\u7ecd\u53ef\u4ee5\u67e5\u9605\u8fd9\u7bc7\u7efc\u8ff0\uff1aImage Segmentation Using Deep Learning: A Survey<\/a>\u3002<\/p>\n

    <\/p>\n

     <\/p>\n","protected":false},"excerpt":{"rendered":"

    \u76ee\u524d\u672c\u4eba\u5728\u6d59\u5927\u8fdb\u4fee\uff0c\u611f\u53d7\u5230AI+VIS\u7684<\/p>\n","protected":false},"author":1,"featured_media":2473,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[12],"tags":[14,152,154],"class_list":["post-2469","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-data-visualization","tag-ai","tag-152","tag-154"],"_links":{"self":[{"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/posts\/2469","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/comments?post=2469"}],"version-history":[{"count":3,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/posts\/2469\/revisions"}],"predecessor-version":[{"id":3074,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/posts\/2469\/revisions\/3074"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/"}],"wp:attachment":[{"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/media?parent=2469"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/categories?post=2469"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/tags?post=2469"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}