{"id":1399,"date":"2024-10-14T22:19:53","date_gmt":"2024-10-14T14:19:53","guid":{"rendered":"https:\/\/www.15zhi.net\/blog\/?p=1399"},"modified":"2024-10-15T14:33:41","modified_gmt":"2024-10-15T06:33:41","slug":"2410-%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-boosting-camouflaged-object-with-dual-task-interactive-transformer","status":"publish","type":"post","link":"https:\/\/www.15zhi.net\/blog\/2410-%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-boosting-camouflaged-object-with-dual-task-interactive-transformer\/","title":{"rendered":"2410\u8bba\u6587\u7814\u8bfb-Boosting Camouflaged Object with Dual-Task Interactive Transformer"},"content":{"rendered":"\n

\u4f5c\u8005\uff1aZhengyi Liu,Zhili Zhang,Yacheng Tan,Wei Wu<\/p>\n\n\n\n

\u8bba\u6587\u6765\u6e90\uff1aInternational Conference on Pattern Recognition (ICPR)<\/p>\n\n\n\n

\u53d1\u8868\u65e5\u671f\uff1a2022 <\/p>\n\n\n\n

\u80cc\u666f<\/h2>\n\n\n\n

\u4f2a\u88c5\u76ee\u6807\u68c0\u6d4b\u7684\u76ee\u6807\u662f\u53d1\u73b0\u9690\u85cf\u5728\u5468\u56f4\u73af\u5883\u4e2d\u7684\u4f2a\u88c5\u7269\u4f53\u3002\u73b0\u6709\u7684\u65b9\u6cd5\u9075\u5faa\u4eff\u751f\u6846\u67b6\uff0c\u5373\u9996\u5148\u5bf9\u7269\u4f53\u8fdb\u884c\u5b9a\u4f4d\uff0c\u7136\u540e\u7ec6\u5316\u7269\u4f53\u7684\u8fb9\u754c\u3002\u4f5c\u8005\u8ba4\u4e3a\uff0c\u4f2a\u88c5\u7269\u4f53\u7684\u53d1\u73b0\u53d6\u51b3\u4e8e\u5bf9\u7269\u4f53\u548c\u8fb9\u754c\u7684\u53cd\u590d\u641c\u7d22\u3002\u53cd\u590d\u7684\u641c\u7d22\u662f\u5177\u6709\u5168\u5c40\u641c\u7d22\u80fd\u529b\u7684Transformer\u6240\u64c5\u957f\u7684\u3002\u4e3a\u6b64\uff0c\u4f5c\u8005\u63d0\u51fa\u4e86\u4e00\u79cd\u53cc\u4efb\u52a1\u4ea4\u4e92\u7684Transformer\u6765\u68c0\u6d4b\u4f2a\u88c5\u76ee\u6807\u7684\u4f4d\u7f6e\u53ca\u5176\u51c6\u786e\u7684\u8fb9\u754c\u3002<\/p>\n\n\n\n

\u521b\u65b0\u70b9<\/h2>\n\n\n\n

\u4f5c\u8005\u63d0\u51fa\u4e86\u4e00\u79cd\u53cc\u4efb\u52a1\u4ea4\u4e92\u7684Transformer\u67b6\u6784\uff0c\u4ee5\u4fc3\u8fdb\u8fb9\u754c\u68c0\u6d4b\u548cCOD\u4efb\u52a1\u4e4b\u95f4\u7684\u4ea4\u4e92\u3002\u4e0e\u5148\u524d\u7684\u4eff\u751f\u6846\u67b6\u4e0d\u540c\uff0c\u4f5c\u8005\u8bbe\u8ba1\u7684\u7f51\u7edc\u6a21\u578b\u662f\u901a\u8fc7\u4ea4\u53c9\u591a\u5934\u6ce8\u610f\u529b\u673a\u5236\u5e76\u884c\u641c\u7d22\u7269\u4f53\u548c\u5176\u8fb9\u754c\uff0c\u8bd5\u56fe\u8d85\u8d8a\u4eba\u7c7b\u7684\u667a\u80fd\uff0c\u800c\u4e0d\u662f\u6a21\u4eff\u4eba\u7c7b\u7684\u89c6\u89c9\u3002<\/p>\n\n\n\n

\u65b9\u6cd5<\/h2>\n\n\n\n

COD\u4efb\u52a1<\/h3>\n\n\n\n

\u4f2a\u88c5\u5bf9\u8c61\u68c0\u6d4b\uff08COD\uff09\u7684\u76ee\u6807\u662f\u5bf9\u9690\u85cf\u5728\u73af\u5883\u4e2d\u7684\u4f2a\u88c5\u7269\u4f53\u8fdb\u884c\u5206\u5272\u3002\u5173\u952e\u7684\u6311\u6218\u5728\u4e8e\uff0c\u4f2a\u88c5\u7684\u7269\u4f53\u5728\u989c\u8272\u3001\u7eb9\u7406\u3001\u4eae\u5ea6\u6216\u5176\u4ed6\u56fe\u6848\u4e0a\u4e0e\u5468\u56f4\u73af\u5883\u76f8\u4f3c\u3002\u7136\u800c\uff0c\u4f2a\u88c5\u5bf9\u8c61\u867d\u7136\u53ef\u4ee5\u9690\u85cf\u5728\u80cc\u666f\u4e2d\uff0c\u4f46\u5b83\u4e0d\u53ef\u907f\u514d\u5730\u6709\u5176\u8fb9\u754c\u3002\u56e0\u6b64\uff0c\u8fb9\u754c\u68c0\u6d4b\u5bf9COD\u4efb\u52a1\u81f3\u5173\u91cd\u8981\u3002<\/p>\n\n\n\n

\u57fa\u4e8e\u4e0a\u8ff0\u7684\u5206\u6790\uff0c\u4f5c\u8005\u63d0\u51fa\u4e86\u63d0\u51fa\u4e86\u4e00\u79cd\u53cc\u4efb\u52a1\u4ea4\u4e92\u5f0fTransformer\u6765\u5e76\u884c\u68c0\u6d4b\u4f2a\u88c5\u7269\u4f53\u548c\u8fb9\u754c\u3002<\/p>\n\n\n\n

\u6a21\u578b\u67b6\u6784<\/h3>\n\n\n\n
\"\"<\/figure>\n\n\n\n

\u4f5c\u8005\u9996\u5148\u901a\u8fc7\u4e24\u4e2a\u5e76\u884c\u7684Transformer\u67b6\u6784\u5206\u522b\u7528\u6765\u63d0\u53d6\u56fe\u7247\u7684\u524d\u666f\u7279\u5f81\u4e0e\u80cc\u666f\u7279\u5f81\u3002\u4ece\u56fe\u4e2d\u53ef\u4ee5\u5f97\u5230\uff0c\u7279\u5f81\u63d0\u53d6\u7684Transformer\u67b6\u6784\u75314\u4e2aTransformer Block\u7ec4\u6210(Transformer Block\u7531Segformer\u7ec4\u6210)\uff0c\u6765\u63d0\u53d6\u4e0d\u540c\u5c3a\u5ea6\u7684\u524d\u666f\u7279\u5f81\u3002\u4f5c\u8005\u8ba4\u4e3a\u7269\u4f53\u7684\u8fb9\u754c\u7279\u5f81\u53ef\u7531\u524d\u666f\u7279\u5f81\u51cf\u53bb\u80cc\u666f\u7279\u5f81\u5f97\u5230\uff0c\u56e0\u6b64\u901a\u8fc7\u7279\u5f81\u63d0\u53d6\u67b6\u6784\u5f97\u5230\u4e864\u4e2a\u4e0d\u540c\u5c3a\u5ea6\u7684\u524d\u666f\u7279\u5f81\u4e0e\u8fb9\u754c\u7279\u5f81\u3002<\/p>\n\n\n\n

\u5c06\u5f97\u5230\u7684\u4e0d\u540c\u5c3a\u5ea6\u7684\u524d\u666f\/\u8fb9\u754c\u7279\u5f81\u901a\u8fc7FEA\u8fdb\u884c\u878d\u5408\uff0c\u518d\u8f93\u5165\u7684\u53cc\u4efb\u52a1\u4ea4\u4e92\u7684Transformer\u67b6\u6784\u4e2d\uff0c\u8f93\u51fa\u6700\u7ec8\u7684\u5206\u5272\u7ed3\u679c\u4e0e\u7269\u4f53\u8fb9\u754c\u9884\u6d4b\u7ed3\u679c\u3002<\/p>\n\n\n\n

FEA<\/h3>\n\n\n\n
\"\"<\/figure>\n\n\n\n

\u5c06\u7279\u5f81\u63d0\u53d6\u6a21\u5757\u8f93\u51fa\u76844\u4e2a\u4e0d\u540c\u5c3a\u5ea6\u7684\u524d\u666f\/\u8fb9\u754c\u7279\u5f81\u8f93\u5165\u5230\u5bf9\u5e94FEA\u4e2d\u8fdb\u884c\u878d\u5408\uff0c\u4ece\u800c\u5f97\u5230\u4e86\u5305\u542b\u5404\u4e2a\u5c3a\u5ea6\u4fe1\u606f\u7684\u524d\u666f\/\u8fb9\u754c\u7279\u5f81\u3002<\/p>\n\n\n\n

Dual-Task Interactive Transformer<\/h3>\n\n\n\n
\"\"<\/figure>\n\n\n\n

\u53cc\u4efb\u52a1\u4ea4\u4e92\u7684Transformer\u67b6\u6784\u75312\u4e2aTransformer\u67b6\u6784\u7ec4\u6210\uff0c\u5206\u522b\u7528\u4e8e\u524d\u666f\u9884\u6d4b\u4efb\u52a1\u548c\u8fb9\u754c\u9884\u6d4b\u4efb\u52a1\u3002\u4ece\u56fe\u4e2d\u53ef\u4ee5\u770b\u51fa\uff0c\u524d\u666f\u7279\u5f81\u5145\u5f53\u5bf9\u5e94\u524d\u666f\u9884\u6d4bTransformer\u7684query,\u800ckey\u548cvalue\u5219\u6765\u81ea\u524d\u666f\u7279\u5f81\u4e0e\u8fb9\u754c\u7279\u5f81\u7684\u62fc\u63a5\u7279\u5f81\u3002\u540c\u6837\u7684\uff0c\u8fb9\u754c\u7279\u5f81\u5145\u5f53\u5bf9\u5e94\u8fb9\u754c\u9884\u6d4bTransformer\u7684query,\u800ckey\u548cvalue\u5219\u6765\u81ea\u8fb9\u754c\u7279\u5f81\u4e0e\u524d\u666f\u7279\u5f81\u7684\u62fc\u63a5\u7279\u5f81\u3002\u4ece\u800c\u5b9e\u73b0\u4e86\uff0c\u5728\u8fdb\u884c\u524d\u666f\u9884\u6d4b\u4efb\u52a1\u65f6\uff0c\u8fb9\u754c\u7684\u7279\u5f81\u63d0\u4f9b\u4f2a\u88c5\u5bf9\u8c61\u7684\u8f6e\u5ed3\u7ec6\u8282\uff0c\u5728\u8fdb\u884c\u8fb9\u754c\u9884\u6d4b\u4efb\u52a1\u65f6\uff0c\u524d\u666f\u7279\u5f81\u5bf9\u4f4d\u7f6e\u7ebf\u7d22\u7684\u8fb9\u754c\u7ec6\u5316\u63d0\u4f9b\u6307\u5bfc\uff0c\u4ece\u800c\u5b9e\u73b0\u4e86\u5e76\u884c\u641c\u7d22\u7269\u4f53\u548c\u5176\u8fb9\u754c\u3002<\/p>\n\n\n\n

\u5b9e\u9a8c<\/h2>\n\n\n\n

Dataset<\/h3>\n\n\n\n

\u5728CAMO,COD10K,NC4K \u7b49\u6570\u636e\u96c6\u4e0a\u8fdb\u884c\u4e86\u5e7f\u6cdb\u7684\u5b9e\u9a8c\u3002<\/p>\n\n\n\n

\u8bc4\u4f30\u6307\u6807<\/h3>\n\n\n\n

\u4f5c\u8005\u91c7\u75285\u4e2a\u5e38\u7528\u7684\u8bc4\u4ef7\u6307\u6807\u6765\u8bc4\u4ef7\u63d0\u51fa\u7684COD\u6a21\u578b\u7684\u6027\u80fd\uff0c\u5305\u62ecprecision-recall \uff08PR\uff09\u66f2\u7ebf\u3001S-measure \uff08S\u03b1<\/sub>\uff09\u3001mean E-measure \uff08E\u03c6<\/sub>\uff09\u3001weighted F-measure (F\u03c9<\/sup>\u03b2<\/sub>)\u548c\u5e73\u5747\u7edd\u5bf9\u8bef\u5dee\uff08MAE\uff09\u3002<\/p>\n\n\n\n

\u5b9e\u9a8c\u7ed3\u679c<\/h3>\n\n\n\n
\"\"<\/figure>\n\n\n\n
\"\"<\/figure>\n\n\n\n

\u901a\u8fc7\u8868\u4e00\uff0c\u53ef\u4ee5\u5f97\u5230\u4f5c\u8005\u7684\u65b9\u6cd5\u5728\u6240\u67094\u4e2a\u8bc4\u4f30\u6307\u6807\u65b9\u9762\u4f18\u4e8e\u6700\u5148\u8fdb\u7684\u6a21\u578b\u3002\u6b64\u5916\uff0c\u4f5c\u8005\u63d0\u51fa\u7684COD\u6a21\u578b\u7684PR\u66f2\u7ebf\u57283\u4e2a\u6570\u636e\u96c6\u4e0a\u5747\u9ad8\u4e8e\u5176\u4ed6\u6a21\u578b\uff0c\u9a8c\u8bc1\u4e86\u6240\u63d0\u51fa\u6a21\u578b\u7684\u6709\u6548\u6027\u3002<\/p>\n\n\n\n

\u603b\u7ed3<\/h2>\n\n\n\n

\u4f5c\u8005\u63d0\u51fa\u4e86\u4e00\u79cd\u5229\u7528\u4ea4\u53c9\u591a\u5934\u81ea\u6ce8\u610f\u673a\u5236\u5e76\u884c\u68c0\u6d4b\u4f2a\u88c5\u76ee\u6807\u548c\u8fb9\u754c\u7684\u65b9\u6cd5\u3002\u5b83\u5145\u5206\u5229\u7528\u4e86\u8ba1\u7b97\u673a\u7684\u5feb\u901f\u5e76\u884c\u8ba1\u7b97\u80fd\u529b\uff0c\u8d85\u8d8a\u4e86\u73b0\u6709\u7684\u4ec5\u6a21\u62df\u4eba\u7c7b\u89c6\u89c9\u7cfb\u7edf\u7684COD\u6a21\u578b\u3002\u8fd9\u79cd\u53cc\u4efb\u52a1\u7684\u4ea4\u4e92\uff0c\u88ab\u9a8c\u8bc1\u662f\u6709\u6548\u7684\uff0c\u53ef\u5c06\u8fd9\u79cd\u4ea4\u4e92\u65b9\u5f0f\u5e94\u7528\u5230\u522b\u7684\u4efb\u52a1\u4e0a\u3002\u6b64\u5916\uff0c\u8fd8\u53ef\u4ee5\u5bf9\u6587\u4e2d\u7684\u7279\u5f81\u63d0\u53d6\u6a21\u5757\u8fdb\u884c\u8c03\u6574\uff0c\u6bd4\u5982\u907f\u514d\u63d0\u53d6\u80cc\u666f\u7279\u5f81\uff0c\u5c06\u5176\u66ff\u6362\u6210\u4e00\u79cd\u66f4\u9ad8\u6548\u7684\u8fb9\u754c\u63d0\u53d6\u65b9\u5f0f\u3002<\/p>\n\n\n\n

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

\u4f5c\u8005\uff1aZhengyi Liu,Zhili Zhang,Yacheng Tan,Wei Wu \u8bba\u6587\u6765\u6e90\uff1aInte […]<\/p>\n","protected":false},"author":7,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-1399","post","type-post","status-publish","format-standard","hentry","category-events"],"_links":{"self":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/1399","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/users\/7"}],"replies":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/comments?post=1399"}],"version-history":[{"count":9,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/1399\/revisions"}],"predecessor-version":[{"id":1465,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/1399\/revisions\/1465"}],"wp:attachment":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/media?parent=1399"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/categories?post=1399"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/tags?post=1399"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}