{"id":36701,"date":"2025-01-16T21:00:16","date_gmt":"2025-01-16T13:00:16","guid":{"rendered":"https:\/\/17aitech.com\/?p=36701"},"modified":"2025-01-16T21:00:16","modified_gmt":"2025-01-16T13:00:16","slug":"neurips-2024-%e6%97%a0%e9%9c%80%e8%ae%ad%e7%bb%83%ef%bc%8c%e4%b8%80%e4%b8%aa%e6%a1%86%e6%9e%b6%e6%90%9e%e5%ae%9a%e5%bc%80%e6%94%be%e5%bc%8f%e7%9b%ae%e6%a0%87%e6%a3%80%e6%b5%8b%e3%80%81%e5%ae%9e","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=36701","title":{"rendered":"NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272"},"content":{"rendered":"<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.jiqizhixin.com\/articles\/2024-11-16-4\" target=\"_blank\">NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272<\/a><\/p>\n<p><strong>\u672c\u6587\u4f5c\u8005\u5747\u6765\u81ea\u5317\u4eac\u5927\u5b66\u738b\u9009\u8ba1\u7b97\u673a\u7814\u7a76\u6240\u3002\u4e3b\u8981\u4f5c\u8005\u5305\u62ec\uff1a\u6797\u5fd7\u5a01\uff0c\u5317\u4eac\u5927\u5b66\u535a\u58eb\u751f\uff1b\u738b\u52c7\u6d9b\uff0c\u5317\u4eac\u5927\u5b66\u526f\u7814\u7a76\u5458\uff1b\u6c64\u5e1c\uff0c\u5317\u4eac\u5927\u5b66\u7814\u7a76\u5458\u3002<\/strong><\/p>\n<p>\u672c\u6587\u4ecb\u7ecd\u4e86\u6765\u81ea\u5317\u4eac\u5927\u5b66\u738b\u9009\u8ba1\u7b97\u673a\u7814\u7a76\u6240\u7684\u738b\u52c7\u6d9b\u56e2\u961f\u7684\u6700\u65b0\u7814\u7a76\u6210\u679c VL-SAM\u3002\u9488\u5bf9\u5f00\u653e\u573a\u666f\uff0c\u8be5\u7bc7\u5de5\u4f5c\u63d0\u51fa\u4e86\u4e00\u4e2a\u57fa\u4e8e\u6ce8\u610f\u529b\u56fe\u63d0\u793a\u7684\u514d\u8bad\u7ec3\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u548c\u5206\u5272\u6846\u67b6 VL-SAM\uff0c\u5728\u65e0\u9700\u8bad\u7ec3\u7684\u60c5\u51b5\u4e0b\uff0c\u53d6\u5f97\u4e86\u826f\u597d\u7684\u5f00\u653e\u5f0f (Open-ended) \u76ee\u6807\u68c0\u6d4b\u548c\u5b9e\u4f8b\u5206\u5272\u7ed3\u679c\uff0c\u8bba\u6587\u5df2\u88ab NeurIPS 2024 \u5f55\u7528\u3002<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-80755e91cc3b3cdd07a168b1a1bc7917.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-80755e91cc3b3cdd07a168b1a1bc7917.png\"><\/a><\/p>\n<ul>\n<li>\u8bba\u6587\u6807\u9898\uff1aTraining-Free Open-Ended Object Detection and Segmentation via Attention as Prompts<\/li>\n<li>\u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/abs\/2410.05963<\/li>\n<\/ul>\n<p><strong>\u8bba\u6587\u6982\u8ff0<\/strong><\/p>\n<p>\u672c\u6587\u63d0\u51fa\u4e86\u4e00\u4e2a\u65e0\u9700\u8bad\u7ec3\u7684\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u548c\u5206\u5272\u6846\u67b6\uff0c\u7ed3\u5408\u4e86\u73b0\u6709\u7684\u6cdb\u5316\u7269\u4f53\u8bc6\u522b\u6a21\u578b\uff08\u5982\u89c6\u89c9\u8bed\u8a00\u5927\u6a21\u578b VLM\uff09\u4e0e\u6cdb\u5316\u7269\u4f53\u5b9a\u4f4d\u6a21\u578b\uff08\u5982\u5206\u5272\u57fa\u7840\u6a21\u578b SAM\uff09\uff0c\u5e76\u4f7f\u7528\u6ce8\u610f\u529b\u56fe\u4f5c\u4e3a\u63d0\u793a\u8fdb\u884c\u4e24\u8005\u7684\u8fde\u63a5\u3002\u5728\u957f\u5c3e\u6570\u636e\u96c6 LVIS \u4e0a\uff0c\u8be5\u6846\u67b6\u8d85\u8fc7\u4e86\u4e4b\u524d\u9700\u8981\u8bad\u7ec3\u7684\u5f00\u653e\u5f0f\u65b9\u6cd5\uff0c\u540c\u65f6\u80fd\u591f\u63d0\u4f9b\u989d\u5916\u7684\u5b9e\u4f8b\u5206\u5272\u7ed3\u679c\u3002\u5728\u81ea\u52a8\u9a7e\u9a76 corner case \u6570\u636e\u96c6 CODA \u4e0a\uff0cVL-SAM \u4e5f\u8868\u73b0\u51fa\u4e86\u4e0d\u9519\u7684\u7ed3\u679c\uff0c\u8bc1\u660e\u4e86\u5176\u5728\u771f\u5b9e\u5e94\u7528\u573a\u666f\u4e0b\u7684\u80fd\u529b\u3002\u6b64\u5916\uff0cVL-SAM \u5c55\u73b0\u4e86\u5f3a\u5927\u7684\u6a21\u578b\u6cdb\u5316\u80fd\u529b\uff0c\u80fd\u591f\u7ed3\u5408\u5f53\u524d\u5404\u79cd VLM \u548c SAM \u6a21\u578b\u3002<\/p>\n<p><strong>\u7814\u7a76\u80cc\u666f<\/strong><\/p>\n<p>\u6df1\u5ea6\u5b66\u4e60\u5728\u611f\u77e5\u4efb\u52a1\u65b9\u9762\u53d6\u5f97\u4e86\u663e\u8457\u6210\u529f\uff0c\u5176\u4e2d\uff0c\u81ea\u52a8\u9a7e\u9a76\u662f\u4e00\u4e2a\u5178\u578b\u7684\u6210\u529f\u6848\u4f8b\u3002\u73b0\u6709\u7684\u57fa\u4e8e\u6df1\u5ea6\u5b66\u4e60\u7684\u611f\u77e5\u6a21\u578b\u4f9d\u8d56\u4e8e\u5e7f\u6cdb\u7684\u6807\u8bb0\u8bad\u7ec3\u6570\u636e\u6765\u5b66\u4e60\u8bc6\u522b\u548c\u5b9a\u4f4d\u5bf9\u8c61\u3002\u7136\u800c\uff0c\u8bad\u7ec3\u6570\u636e\u4e0d\u80fd\u5b8c\u5168\u8986\u76d6\u771f\u5b9e\u4e16\u754c\u573a\u666f\u4e2d\u6240\u6709\u7c7b\u578b\u7684\u7269\u4f53\u3002\u5f53\u9762\u5bf9\u5206\u5e03\u5916\u7684\u7269\u4f53\u65f6\uff0c\u73b0\u6709\u7684\u611f\u77e5\u6a21\u578b\u53ef\u80fd\u65e0\u6cd5\u8fdb\u884c\u8bc6\u522b\u548c\u5b9a\u4f4d\uff0c\u4ece\u800c\u53ef\u80fd\u4f1a\u53d1\u751f\u4e25\u91cd\u7684\u5b89\u5168\u95ee\u9898\u3002<\/p>\n<p>\u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\uff0c\u7814\u7a76\u8005\u4eec\u63d0\u51fa\u4e86\u8bb8\u591a\u5f00\u653e\u4e16\u754c\u611f\u77e5\u65b9\u6cd5\u3002\u8fd9\u4e9b\u65b9\u6cd5\u5927\u81f4\u53ef\u4ee5\u5206\u4e3a\u4e24\u7c7b\uff1a\u5f00\u96c6\u611f\u77e5\uff08open-set\uff09\u548c\u5f00\u653e\u5f0f\u611f\u77e5\uff08open-ended\uff09\u3002\u5f00\u96c6\u611f\u77e5\u65b9\u6cd5\u901a\u5e38\u4f7f\u7528\u9884\u8bad\u7ec3\u7684 CLIP \u6a21\u578b\u6765\u8ba1\u7b97\u56fe\u50cf\u533a\u57df\u548c\u7c7b\u522b\u540d\u79f0\u4e4b\u95f4\u7684\u76f8\u4f3c\u6027\u3002\u56e0\u6b64\uff0c\u5728\u63a8\u7406\u8fc7\u7a0b\u4e2d\uff0c\u8fd9\u7c7b\u65b9\u6cd5\u9700\u8981\u9884\u5b9a\u4e49\u7684\u5bf9\u8c61\u7c7b\u522b\u540d\u79f0\u4f5c\u4e3a CLIP \u6587\u672c\u7f16\u7801\u5668\u7684\u8f93\u5165\u3002\u7136\u800c\uff0c\u5728\u8bb8\u591a\u73b0\u5b9e\u4e16\u754c\u7684\u5e94\u7528\u573a\u666f\u4e2d\uff0c\u5e76\u4e0d\u4f1a\u63d0\u4f9b\u786e\u5207\u7684\u5bf9\u8c61\u7c7b\u522b\u540d\u79f0\u3002\u4f8b\u5982\uff0c\u5728\u81ea\u52a8\u9a7e\u9a76\u573a\u666f\u4e2d\uff0c\u81ea\u52a8\u9a7e\u9a76\u8f66\u8f86\u53ef\u80fd\u4f1a\u9047\u5230\u5404\u79cd\u610f\u60f3\u4e0d\u5230\u7684\u7269\u4f53\uff0c\u5305\u62ec\u8d77\u706b\u6216\u4fa7\u7ffb\u7684\u4e8b\u6545\u8f66\u548c\u5404\u79cd\u5404\u6837\u7684\u5efa\u7b51\u8f66\u8f86\u3002\u76f8\u6bd4\u4e4b\u4e0b\uff0c\u5f00\u653e\u5f0f\u611f\u77e5\u65b9\u6cd5\u66f4\u5177\u901a\u7528\u6027\u548c\u5b9e\u7528\u6027\uff0c\u56e0\u4e3a\u8fd9\u4e9b\u53ef\u4ee5\u540c\u65f6\u9884\u6d4b\u5bf9\u8c61\u7c7b\u522b\u548c\u4f4d\u7f6e\uff0c\u800c\u4e0d\u9700\u8981\u7ed9\u5b9a\u786e\u5207\u7684\u5bf9\u8c61\u7c7b\u522b\u540d\u79f0\u3002\u00a0<\/p>\n<p>\u4e0e\u6b64\u540c\u65f6\uff0c\u5728\u6700\u8fd1\u7684\u7814\u7a76\u4e2d\uff0c\u5927\u578b\u89c6\u89c9\u8bed\u8a00\u6a21\u578b\uff08VLM\uff09\u663e\u793a\u51fa\u5f3a\u5927\u7684\u7269\u4f53\u8bc6\u522b\u6cdb\u5316\u80fd\u529b\uff0c\u4f8b\u5982\uff0c\u5b83\u53ef\u4ee5\u5728\u81ea\u52a8\u9a7e\u9a76\u573a\u666f\u4e2d\u7684\u957f\u5c3e\u6570\u636e\u4e0a\uff08corner case\uff09\u8bc6\u522b\u975e\u5e38\u89c1\u7684\u7269\u4f53\uff0c\u5e76\u7ed9\u51fa\u51c6\u786e\u7684\u63cf\u8ff0\u3002\u7136\u800c\uff0cVLM \u7684\u5b9a\u4f4d\u80fd\u529b\u76f8\u6bd4\u4e8e\u7279\u5b9a\u611f\u77e5\u6a21\u578b\u8f83\u5f31\uff0c\u7ecf\u5e38\u4f1a\u6f0f\u68c0\u7269\u4f53\u6216\u7ed9\u51fa\u9519\u8bef\u7684\u5b9a\u4f4d\u7ed3\u679c\u3002\u53e6\u4e00\u65b9\u9762\uff0c\u4f5c\u4e3a\u4e00\u4e2a\u7eaf\u89c6\u89c9\u57fa\u7840\u6a21\u578b\uff0cSAM \u5bf9\u6765\u81ea\u8bb8\u591a\u4e0d\u540c\u9886\u57df\u7684\u56fe\u50cf\u8868\u73b0\u51fa\u826f\u597d\u7684\u5206\u5272\u6cdb\u5316\u80fd\u529b\u3002\u7136\u800c\uff0cSAM \u65e0\u6cd5\u4e3a\u5206\u5272\u7684\u5bf9\u8c61\u63d0\u4f9b\u7c7b\u522b\u3002\u57fa\u4e8e\u6b64\uff0c\u672c\u6587\u63d0\u51fa\u4e86\u4e00\u4e2a\u65e0\u9700\u8bad\u7ec3\u7684\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u548c\u5206\u5272\u6846\u67b6 VL-SAM\uff0c\u5c06\u73b0\u6709\u7684\u6cdb\u5316\u7269\u4f53\u8bc6\u522b\u6a21\u578b VLM \u4e0e\u6cdb\u5316\u7269\u4f53\u5b9a\u4f4d\u6a21\u578b SAM \u76f8\u7ed3\u5408\uff0c\u5229\u7528\u6ce8\u610f\u529b\u56fe\u4f5c\u4e3a\u4e2d\u95f4\u63d0\u793a\u8fdb\u884c\u8fde\u63a5\uff0c\u4ee5\u89e3\u51b3\u5f00\u653e\u5f0f\u611f\u77e5\u4efb\u52a1\u3002<\/p>\n<p><strong>\u65b9\u6cd5\u90e8\u5206<\/strong><\/p>\n<p>\u4f5c\u8005\u63d0\u51fa\u4e86 VL-SAM\uff0c\u4e00\u4e2a\u65e0\u9700\u8bad\u7ec3\u7684\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u548c\u5206\u5272\u6846\u67b6\u3002\u5177\u4f53\u6846\u67b6\u5982\u4e0b\u56fe\u6240\u793a\uff1a<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-45d110d3c3f92ddfd2a6eb50e86cf998.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-45d110d3c3f92ddfd2a6eb50e86cf998.png\"><\/a><\/p>\n<p><em>\u56fe 1 VL-SAM \u6846\u67b6\u56fe<\/em><\/p>\n<p>\u5177\u4f53\u800c\u8a00\uff0c\u4f5c\u8005\u8bbe\u8ba1\u4e86\u6ce8\u610f\u529b\u56fe\u751f\u6210\u6a21\u5757\uff0c\u91c7\u7528\u5934\u805a\u5408\u548c\u6ce8\u610f\u529b\u6d41\u7684\u65b9\u5f0f\u5bf9\u591a\u5c42\u591a\u5934\u6ce8\u610f\u529b\u56fe\u8fdb\u884c\u4f20\u64ad\uff0c\u4ece\u800c\u751f\u6210\u9ad8\u8d28\u91cf\u7684\u6ce8\u610f\u529b\u56fe\u3002\u4e4b\u540e\uff0c\u4f5c\u8005\u4f7f\u7528\u8fed\u4ee3\u5f0f\u6b63\u8d1f\u6837\u672c\u70b9\u91c7\u6837\u7684\u65b9\u5f0f\uff0c\u4ece\u751f\u6210\u7684\u6ce8\u610f\u529b\u56fe\u4e2d\u8fdb\u884c\u91c7\u6837\uff0c\u5f97\u5230 SAM \u7684\u70b9\u63d0\u793a\u4f5c\u4e3a\u8f93\u5165\uff0c\u6700\u7ec8\u5f97\u5230\u7269\u4f53\u7684\u5206\u5272\u7ed3\u679c\u3002<\/p>\n<p><strong>1\u3001\u6ce8\u610f\u529b\u56fe\u751f\u6210\u6a21\u5757\uff08Attention Map Generation Module\uff09<\/strong><\/p>\n<p>\u7ed9\u5b9a\u4e00\u5f20\u8f93\u5165\u56fe\u7247\uff0c\u4f7f\u7528 VLM \u7ed9\u51fa\u56fe\u7247\u4e2d\u6240\u6709\u7684\u7269\u4f53\u7c7b\u522b\u3002\u5728\u8fd9\u4e2a\u8fc7\u7a0b\u4e2d\u5b58\u50a8 VLM \u751f\u6210\u7684\u6240\u6709 query \u548c key\uff0c\u5e76\u4f7f\u7528 query \u548c key \u6784\u5efa\u591a\u5c42\u591a\u5934\u6ce8\u610f\u529b\u56fe\uff1a<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-0ff22b1594c45fbbcd0a52332ac15ae2.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-0ff22b1594c45fbbcd0a52332ac15ae2.png\"><\/a><\/p>\n<p>\u5176\u4e2d N \u8868\u793a token \u7684\u6570\u91cf\uff0cH \u8868\u793a\u591a\u5934\u6ce8\u610f\u529b\u7684\u6570\u91cf\uff0cL \u8868\u793a VLM \u7684\u5c42\u6570\u3002<\/p>\n<p>\u4e4b\u540e\uff0c\u91c7\u7528 Mean-max \u7684\u65b9\u5f0f\u5bf9\u591a\u5934\u6ce8\u610f\u529b\u56fe\u8fdb\u884c\u805a\u5408\uff0c\u5982\u56fe 2 \u6240\u793a\uff1a<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-4bab41b6b4ad85f2158a669a5ae490d3.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-4bab41b6b4ad85f2158a669a5ae490d3.png\"><\/a><\/p>\n<p><em>\u56fe 2 \u591a\u5934\u6ce8\u610f\u529b\u805a\u5408<\/em><\/p>\n<p>\u9996\u5148\u8ba1\u7b97\u6bcf\u4e2a\u5934\u7684\u6ce8\u610f\u529b\u7684\u6743\u91cd\uff1a<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-435412904fd8383efa54728ea444f503.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-435412904fd8383efa54728ea444f503.png\"><\/a><\/p>\n<p>\u4e4b\u540e\u91c7\u7528\u57fa\u4e8e\u6743\u91cd\u7684\u591a\u5934\u6ce8\u610f\u529b\u52a0\u6743\u8fdb\u884c\u4fe1\u606f\u805a\u5408\uff1a<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-f8dbbfef5834705dae607574ff5a6ebf.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-f8dbbfef5834705dae607574ff5a6ebf.png\"><\/a><\/p>\n<p>\u5176\u4e2d<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-f1149d115d13336987a4604a8e9db191.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-f1149d115d13336987a4604a8e9db191.png\"><\/a>\u8868\u793a\u77e9\u9635\u70b9\u4e58\u3002<\/p>\n<p>\u5728\u805a\u5408\u591a\u5934\u6ce8\u610f\u529b\u56fe\u4e4b\u540e\uff0c\u91c7\u7528\u6ce8\u610f\u529b\u6d41\u7684\u65b9\u5f0f\u8fdb\u4e00\u6b65\u805a\u5408\u591a\u5c42\u6ce8\u610f\u529b\u56fe\uff0c\u5982\u56fe 3 \u6240\u793a<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-931f0baa3a99db5007f2ad9971a85a33.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-931f0baa3a99db5007f2ad9971a85a33.png\"><\/a><\/p>\n<p><em>\u56fe 3 \u6ce8\u610f\u529b\u6d41<\/em><\/p>\n<p>\u5177\u4f53\u800c\u8a00\uff0c\u91c7\u7528 attention rollout \u7684\u65b9\u5f0f\uff0c\u8ba1\u7b97\u7b2c<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-fe95bb4a170c59b21bbffd37d2005e28.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-fe95bb4a170c59b21bbffd37d2005e28.png\"><\/a>\u5c42\u5230\u7b2c<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-41a736eb33470008be95045eb9ae48fc.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-41a736eb33470008be95045eb9ae48fc.png\"><\/a>\u5c42\u7684\u6ce8\u610f\u529b\u56fe\u4f20\u64ad\uff1a<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-fe8d2559f4757b9e297452bac86ecca0.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-fe8d2559f4757b9e297452bac86ecca0.png\"><\/a><\/p>\n<p>\u5176\u4e2d<a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-42159b47d0dcce0e97c1ef9acd541295.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-42159b47d0dcce0e97c1ef9acd541295.png\"><\/a>\u8868\u793a\u5355\u4f4d\u77e9\u9635\u3002\u6700\u540e\uff0c\u4f5c\u8005\u4ec5\u4f7f\u7528\u4f20\u64ad\u540e\u7684\u6700\u540e\u4e00\u5c42\u6ce8\u610f\u529b\u56fe\u4f5c\u4e3a\u6700\u7ec8\u7684\u6ce8\u610f\u529b\u56fe\u3002<\/p>\n<p><strong>2\u3001SAM \u63d0\u793a\u751f\u6210<\/strong><\/p>\n<p>\u751f\u6210\u7684\u6ce8\u610f\u529b\u56fe\u4e2d\u53ef\u80fd\u4f1a\u5b58\u5728\u4e0d\u7a33\u5b9a\u7684\u5047\u9633\u6027\u5cf0\u503c\u3002\u4e3a\u4e86\u8fc7\u6ee4\u8fd9\u90e8\u5206\u5047\u9633\u6027\uff0c\u4f5c\u8005\u9996\u5148\u91c7\u7528\u9608\u503c\u8fc7\u6ee4\u7684\u65b9\u5f0f\u8fdb\u884c\u521d\u6b65\u8fc7\u6ee4\uff0c\u5e76\u627e\u5230\u5269\u4f59\u6fc0\u6d3b\u90e8\u5206\u7684\u6700\u5927\u8054\u901a\u533a\u57df\u4f5c\u4e3a\u6b63\u6837\u672c\u533a\u57df\uff0c\u5176\u4f59\u7684\u90e8\u5206\u4f5c\u4e3a\u8d1f\u6837\u672c\u533a\u57df\u3002\u4e4b\u540e\uff0c\u91c7\u7528\u5cf0\u503c\u68c0\u6d4b\u7684\u65b9\u5f0f\u5206\u522b\u4ece\u6b63\u8d1f\u6837\u672c\u533a\u57df\u8fdb\u884c\u91c7\u6837\uff0c\u5f97\u5230\u6b63\u8d1f\u6837\u672c\u70b9\uff0c\u4f5c\u4e3a SAM \u7684\u70b9\u63d0\u793a\u8f93\u5165\u3002<\/p>\n<p><strong>3\u3001\u8fed\u4ee3\u5f0f\u5206\u5272\u4f18\u5316<\/strong><\/p>\n<p>\u4ece SAM \u5f97\u5230\u5206\u5272\u7ed3\u679c\u53ef\u80fd\u4f1a\u5b58\u5728\u7c97\u7cd9\u7684\u8fb9\u754c\u6216\u8005\u80cc\u666f\u566a\u58f0\uff0c\u4f5c\u8005\u91c7\u7528\u4e24\u79cd\u8fed\u4ee3\u5f0f\u65b9\u5f0f\u8fdb\u4e00\u6b65\u5bf9\u5206\u5272\u7ed3\u679c\u8fdb\u884c\u4f18\u5316\u3002\u5728\u7b2c\u4e00\u79cd\u8fed\u4ee3\u65b9\u5f0f\u4e2d\uff0c\u4f5c\u8005\u501f\u9274 PerSAM \u4f7f\u7528 cascaded post-refinement \u7684\u65b9\u5f0f\uff0c\u5c06\u521d\u59cb\u7684\u5206\u5272\u7ed3\u679c\u4f5c\u4e3a\u989d\u5916\u7684\u63d0\u793a\u8f93\u5165\u5230 SAM \u4e2d\u3002\u5bf9\u4e8e\u7b2c\u4e8c\u79cd\u8fed\u4ee3\u65b9\u5f0f\uff0c\u4f5c\u8005\u4f7f\u7528\u521d\u59cb\u7684\u5206\u5272\u7ed3\u679c\u5bf9\u6ce8\u610f\u529b\u56fe\u8fdb\u884c\u63a9\u7801\uff0c\u4e4b\u540e\u5728\u63a9\u7801\u7684\u533a\u57df\u8fdb\u884c\u6b63\u8d1f\u6837\u672c\u70b9\u91c7\u6837\u3002<\/p>\n<p><strong>4\u3001\u591a\u5c3a\u5ea6\u805a\u5408\u548c\u95ee\u9898\u63d0\u793a\u805a\u5408<\/strong><\/p>\n<p>\u4f5c\u8005\u8fd8\u91c7\u7528\u4e24\u79cd\u805a\u5408\uff08Ensemble\uff09\u7684\u65b9\u5f0f\u8fdb\u4e00\u6b65\u6539\u826f\u7ed3\u679c\u3002\u5bf9\u4e8e VLM \u7684\u4f4e\u5206\u7387\u95ee\u9898\uff0c\u4f5c\u8005\u4f7f\u7528\u591a\u5c3a\u5ea6\u805a\u5408\uff0c\u5c06\u56fe\u7247\u5207\u6210 4 \u5757\u8fdb\u884c\u8f93\u5165\u3002\u6b64\u5916\uff0c\u7531\u4e8e VLM \u5bf9\u95ee\u9898\u8f93\u5165\u8f83\u4e3a\u654f\u611f\uff0c\u4f5c\u8005\u91c7\u7528\u95ee\u9898\u63d0\u793a\u805a\u5408\uff0c\u4f7f\u5f97 VLM \u80fd\u591f\u5c3d\u91cf\u591a\u5f97\u8f93\u51fa\u7269\u4f53\u7c7b\u522b\u3002\u6700\u540e\uff0c\u91c7\u7528 NMS \u5bf9\u8fd9\u4e9b\u805a\u5408\u7ed3\u679c\u8fdb\u884c\u8fc7\u6ee4\u3002<\/p>\n<p><strong>\u5b9e\u9a8c\u7ed3\u679c<\/strong><\/p>\n<p>\u5728\u5305\u542b 1203 \u7c7b\u7269\u4f53\u7c7b\u522b\u7684\u957f\u5c3e\u6570\u636e\u96c6 LVIS \u9a8c\u8bc1\u96c6\u4e0a\uff0c\u76f8\u6bd4\u4e8e\u4e4b\u524d\u7684\u5f00\u653e\u5f0f\u65b9\u6cd5\uff0cVL-SAM \u53d6\u5f97\u4e86\u66f4\u9ad8\u7684\u5305\u56f4\u6846 AP \u503c\u3002\u540c\u65f6\uff0cVL-SAM \u8fd8\u80fd\u591f\u83b7\u53d6\u7269\u4f53\u5206\u5272\u7ed3\u679c\u3002\u6b64\u5916\uff0c\u76f8\u6bd4\u4e8e\u5f00\u96c6\u68c0\u6d4b\u65b9\u6cd5\uff0cVL-SAM \u4e5f\u53d6\u5f97\u4e86\u5177\u6709\u7ade\u4e89\u529b\u7684\u6027\u80fd\u3002\u00a0<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-3605aed0cf8bc476c77f9767617415ce.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-3605aed0cf8bc476c77f9767617415ce.png\"><\/a><\/p>\n<p><em>\u8868 1 LVIS \u7ed3\u679c<\/em><\/p>\n<p>\u5728\u81ea\u52a8\u9a7e\u9a76\u573a\u666f corner case \u6570\u636e\u96c6 CODA \u4e0a\uff0cVL-SAM \u4e5f\u53d6\u5f97\u4e86\u4e0d\u9519\u7684\u7ed3\u679c\uff0c\u8d85\u8fc7\u4e86\u5f00\u96c6\u68c0\u6d4b\u548c\u5f00\u653e\u5f0f\u68c0\u6d4b\u7684\u65b9\u6cd5\u3002<\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-f613583ca2f8b2090a0fff9813ee7283.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-f613583ca2f8b2090a0fff9813ee7283.png\"><\/a><\/p>\n<p><em>\u8868 2 CODA \u7ed3\u679c<\/em><\/p>\n<p><strong>\u7ed3\u8bba<\/strong><\/p>\n<p>\u672c\u6587\u63d0\u51fa\u4e86 VL-SAM\uff0c\u4e00\u4e2a\u57fa\u4e8e\u6ce8\u610f\u529b\u56fe\u63d0\u793a\u7684\u514d\u8bad\u7ec3\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u548c\u5206\u5272\u6846\u67b6 VL-SAM\uff0c\u5728\u65e0\u9700\u8bad\u7ec3\u7684\u60c5\u51b5\u4e0b\uff0c\u53d6\u5f97\u4e86\u826f\u597d\u7684\u5f00\u653e\u5f0f (Open-ended) \u76ee\u6807\u68c0\u6d4b\u548c\u5b9e\u4f8b\u5206\u5272\u7ed3\u679c\u3002<\/p>\n<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.jiqizhixin.com\/articles\/2024-11-16-4\" target=\"_blank\">NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:NeurIPS 202 [&hellip;]<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"","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":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","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":[27],"tags":[68,65],"class_list":["post-36701","post","type-post","status-publish","format-standard","hentry","category-news","tag-68","tag-65"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.4 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272 - \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=36701\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/17aitech.com\/?p=36701\",\"url\":\"https:\/\/17aitech.com\/?p=36701\",\"name\":\"NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272 - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\/\/17aitech.com\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/17aitech.com\/?p=36701#primaryimage\"},\"image\":{\"@id\":\"https:\/\/17aitech.com\/?p=36701#primaryimage\"},\"thumbnailUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-80755e91cc3b3cdd07a168b1a1bc7917.png\",\"datePublished\":\"2025-01-16T13:00:16+00:00\",\"author\":{\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\"},\"breadcrumb\":{\"@id\":\"https:\/\/17aitech.com\/?p=36701#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/17aitech.com\/?p=36701\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/?p=36701#primaryimage\",\"url\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-80755e91cc3b3cdd07a168b1a1bc7917.png\",\"contentUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-80755e91cc3b3cdd07a168b1a1bc7917.png\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/17aitech.com\/?p=36701#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/17aitech.com\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272\"}]},{\"@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\/60225458499e817ae0af73e67e440b9d\",\"name\":\"AI\u5c0f\u52a9\u624b\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/image\/\",\"url\":\"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png\",\"contentUrl\":\"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png\",\"caption\":\"AI\u5c0f\u52a9\u624b\"},\"description\":\"\u8fd9\u4e2a\u4eba\u5f88\u61d2\uff0c\u4ec0\u4e48\u90fd\u6ca1\u6709\u7559\u4e0b\uff5e\",\"url\":\"https:\/\/17aitech.com\/?page_id=33738&user=3\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272 - \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=36701","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=36701","url":"https:\/\/17aitech.com\/?p=36701","name":"NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272 - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=36701#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=36701#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-80755e91cc3b3cdd07a168b1a1bc7917.png","datePublished":"2025-01-16T13:00:16+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=36701#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=36701"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=36701#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-80755e91cc3b3cdd07a168b1a1bc7917.png","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/11\/frc-80755e91cc3b3cdd07a168b1a1bc7917.png"},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=36701#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"NeurIPS 2024 | \u65e0\u9700\u8bad\u7ec3\uff0c\u4e00\u4e2a\u6846\u67b6\u641e\u5b9a\u5f00\u653e\u5f0f\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272"}]},{"@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\/60225458499e817ae0af73e67e440b9d","name":"AI\u5c0f\u52a9\u624b","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/#\/schema\/person\/image\/","url":"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png","contentUrl":"\/\/17aitech.com\/wp-content\/uploads\/2024\/04\/robot_3.png","caption":"AI\u5c0f\u52a9\u624b"},"description":"\u8fd9\u4e2a\u4eba\u5f88\u61d2\uff0c\u4ec0\u4e48\u90fd\u6ca1\u6709\u7559\u4e0b\uff5e","url":"https:\/\/17aitech.com\/?page_id=33738&user=3"}]}},"_links":{"self":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/36701","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\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=36701"}],"version-history":[{"count":0,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/36701\/revisions"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=36701"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=36701"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=36701"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}