{"id":26632,"date":"2025-03-24T17:24:30","date_gmt":"2025-03-24T09:24:30","guid":{"rendered":"http:\/\/192.168.10.115\/?p=26632"},"modified":"2025-03-24T17:24:30","modified_gmt":"2025-03-24T09:24:30","slug":"2025-03-24-%e5%8d%97%e5%a4%a7%e5%b0%8f%e7%b1%b3%e6%9c%80%e6%96%b0%e5%b7%a5%e4%bd%9c%ef%bc%81mila%ef%bc%9a%e7%94%a8%e4%ba%8e%e8%87%aa%e5%8a%a8%e9%a9%be%e9%a9%b6%e7%9a%84%e7%8e%af%e8%a7%86%e8%a7%86","status":"publish","type":"post","link":"http:\/\/222.128.65.89:18086\/index.php\/2025\/03\/24\/26632\/","title":{"rendered":"2025-03-24 \u5357\u5927&amp;\u5c0f\u7c73\u6700\u65b0\u5de5\u4f5c\uff01MiLA\uff1a\u7528\u4e8e\u81ea\u52a8\u9a7e\u9a76\u7684\u73af\u89c6\u89c6\u9891\u751f\u6210\u4e16\u754c\u6a21\u578b\uff0cSOTA"},"content":{"rendered":"\n<p>\u539f\u6587\u94fe\u63a5\uff1a<a href=\"https:\/\/zhuanlan.zhihu.com\/p\/32273152617\">https:\/\/zhuanlan.zhihu.com\/p\/32273152617<\/a><\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"h_32273152617_0\"><a href=\"https:\/\/zhida.zhihu.com\/search?content_id=255476962&amp;content_type=Article&amp;match_order=1&amp;q=MiLA&amp;zhida_source=entity\" target=\"_blank\" rel=\"noreferrer noopener\">MiLA<\/a><\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8bba\u6587\u6807\u9898\uff1aMiLA: Multi-view Intensive-fidelity Long-term Video Generation World Model for Autonomous Driving<\/li>\n\n\n\n<li>\u8bba\u6587\u94fe\u63a5\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/arxiv.org\/abs\/2503.15875\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/arxiv.org\/abs\/2503.15875<\/a><\/li>\n\n\n\n<li>\u9879\u76ee\u4e3b\u9875\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/github.com\/xiaomi-mlab\/mila.github.io\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/github.com\/xiaomi-mlab\/mila.github.io<\/a><\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/pic3.zhimg.com\/v2-82c93415b9d45a6616cc49c978ee6cb4_1440w.jpg\" alt=\"\"\/><\/figure>\n\n\n\n<p><strong>\u6838\u5fc3\u521b\u65b0\u70b9\uff1a<\/strong><\/p>\n\n\n\n<p><strong>1. Coarse-to-(Re)fine \u5206\u5c42\u751f\u6210\u6846\u67b6<\/strong><\/p>\n\n\n\n<p>\u63d0\u51fa\u4e24\u9636\u6bb5\u89c6\u9891\u751f\u6210\u6d41\u7a0b\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Coarse\u9636\u6bb5<\/strong>\uff1a\u751f\u6210\u4f4e\u5e27\u7387\u951a\u5e27\uff08Anchor Frames\uff09\uff0c\u4f5c\u4e3a\u5168\u5c40\u573a\u666f\u7ed3\u6784\u57fa\u51c6\uff1b<\/li>\n\n\n\n<li><strong>Refine\u9636\u6bb5<\/strong>\uff1a\u57fa\u4e8e\u951a\u5e27\u901a\u8fc7\u63d2\u503c\u751f\u6210\u9ad8\u5e27\u7387\u89c6\u9891\uff0c\u540c\u65f6\u5f15\u5165<strong>\u8054\u5408\u53bb\u566a\u4e0e\u6821\u6b63\u6d41\uff08JDC\uff09<\/strong>\uff0c\u5728\u63d2\u503c\u8fc7\u7a0b\u4e2d\u540c\u6b65\u4f18\u5316\u951a\u5e27\u7684\u9759\u6001\u7ed3\u6784\u4fdd\u771f\u5ea6\u4e0e\u52a8\u6001\u7269\u4f53\u8fde\u7eed\u6027\uff0c\u6709\u6548\u6291\u5236\u7d2f\u79ef\u8bef\u5dee\u4e0e\u8fd0\u52a8\u5931\u771f\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>2. \u8054\u5408\u53bb\u566a\u4e0e\u6821\u6b63\u6d41\uff08Joint Denoising and Correcting Flow, JDC\uff09<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u901a\u8fc7\u566a\u58f0\u5206\u89e3\u4e0e\u91cd\u53c2\u6570\u5316\u6280\u672f\uff0c\u5c06\u951a\u5e27\u566a\u58f0\u5efa\u6a21\u4e3a<strong>\u7ed3\u6784\u5316\u566a\u58f0\uff08\u9759\u6001\u80cc\u666f\u8bef\u5dee\uff09\u4e0e\u968f\u673a\u566a\u58f0\uff08\u52a8\u6001\u7269\u4f53\u5931\u771f\uff09<\/strong>\u7684\u6df7\u5408\u5206\u5e03\uff1b<\/li>\n\n\n\n<li>\u8bbe\u8ba1\u52a8\u6001\u566a\u58f0\u8c03\u5ea6\u7b56\u7565\uff0c\u8054\u5408\u4f18\u5316\u4f4e\u5e27\u7387\u951a\u5e27\u7684\u6821\u6b63\u4e0e\u9ad8\u5e27\u7387\u63d2\u503c\u5e27\u7684\u5e73\u6ed1\u6027\uff0c\u663e\u8457\u63d0\u5347\u957f\u89c6\u9891\u7684\u52a8\u6001\u8fde\u8d2f\u6027\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>3. \u65f6\u5e8f\u6e10\u8fdb\u53bb\u566a\u8c03\u5ea6\u5668\uff08Temporal Progressive Denoising Scheduler, TPD\uff09<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u63d0\u51fa\u57fa\u4e8e\u4f59\u5f26\u9000\u706b\u51fd\u6570\u7684\u975e\u5747\u5300\u53bb\u566a\u7b56\u7565\uff0c\u5bf9\u65f6\u95f4\u90bb\u8fd1\u5e27\u65bd\u52a0\u5f3a\u65f6\u5e8f\u4f9d\u8d56\u7ea6\u675f\uff0c\u4f18\u5148\u53bb\u566a\u9760\u8fd1\u6761\u4ef6\u5e27\u7684\u7247\u6bb5\uff1b<\/li>\n\n\n\n<li>\u901a\u8fc7\u201c\u5feb-\u6162\u201d\u6a21\u5f0f\u4f18\u5316\u53bb\u566a\u8f68\u8ff9\uff0c\u65e9\u671f\u5e27\u5feb\u901f\u6536\u655b\u4ee5\u63d0\u4f9b\u4e0a\u4e0b\u6587\uff0c\u540e\u671f\u5e27\u7cbe\u7ec6\u5316\u8c03\u6574\uff0c\u6700\u7ec8\u5b9e\u73b0\u5168\u5e8f\u5217\u540c\u6b65\u9ad8\u8d28\u91cf\u751f\u6210\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>4. \u591a\u6a21\u6001\u878d\u5408\u4e0e\u591a\u89c6\u89d2\u5bf9\u9f50\u673a\u5236<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u96c6\u6210\u76f8\u673a\u53c2\u6570\u3001\u8def\u5f84\u70b9\uff08Waypoints\uff09\u4e0e\u6587\u672c\u63cf\u8ff0\u7b49\u591a\u6a21\u6001\u6761\u4ef6\uff0c\u901a\u8fc7\u5085\u91cc\u53f6\u5d4c\u5165\u4e0eMLP\u5b9e\u73b0\u8de8\u6a21\u6001\u7279\u5f81\u878d\u5408\uff1b<\/li>\n\n\n\n<li>\u8bbe\u8ba1\u591a\u89c6\u56fe\u589e\u5f3a\u7a7a\u95f4\u6ce8\u610f\u529b\uff0c\u6269\u5c55DiT\u6a21\u5757\u7684\u8de8\u89c6\u56fe\u611f\u53d7\u91ce\uff0c\u786e\u4fdd\u591a\u89c6\u89d2\u89c6\u9891\u7684\u7a7a\u95f4\u5bf9\u9f50\u4e0e\u51e0\u4f55\u4e00\u81f4\u6027\u3002<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"h_32273152617_1\"><a href=\"https:\/\/zhida.zhihu.com\/search?content_id=255476962&amp;content_type=Article&amp;match_order=1&amp;q=BadToken&amp;zhida_source=entity\" target=\"_blank\" rel=\"noreferrer noopener\">BadToken<\/a><\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8bba\u6587\u6807\u9898\uff1aBadToken: Token-level Backdoor Attacks to Multi-modal Large Language Models<\/li>\n\n\n\n<li>\u8bba\u6587\u94fe\u63a5\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/arxiv.org\/abs\/2503.16023\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/arxiv.org\/abs\/2503.16023<\/a><\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/pic3.zhimg.com\/v2-87889b9fc0ae750b8fa88e94b9bdd18c_1440w.jpg\" alt=\"\"\/><\/figure>\n\n\n\n<p><strong>\u6838\u5fc3\u521b\u65b0\u70b9\uff1a<\/strong><\/p>\n\n\n\n<p><strong>1. \u53cc\u6a21\u5f0f\u540e\u95e8\u884c\u4e3a\u5b9a\u4e49<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Token-substitution \uff1a\u901a\u8fc7\u66ff\u6362\u8f93\u51fa\u5e8f\u5217\u4e2d\u7684\u7279\u5b9a\u6e90\u8bcd\u5143\uff08\u5982&#8221;red&#8221;\u2192&#8221;green&#8221;\uff09\uff0c\u5b9e\u73b0\u8bed\u4e49\u7ffb\u8f6c\u7684\u9690\u853d\u653b\u51fb<\/li>\n\n\n\n<li>Token-addition \uff1a\u5728\u8f93\u51fa\u5c3e\u90e8\u6ce8\u5165\u9884\u8bbe\u6076\u610f\u8bcd\u5143\u5e8f\u5217\uff08\u5982\u6076\u610fURL\uff09\uff0c\u4fdd\u6301\u8bed\u4e49\u5b8c\u6574\u6027\u7684\u540c\u65f6\u5d4c\u5165\u6709\u5bb3\u4fe1\u606f<\/li>\n<\/ul>\n\n\n\n<p><strong>2. \u4f18\u5316\u76ee\u6807\u5efa\u6a21<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u63d0\u51fa<strong>\u6709\u6548\u6027\u635f\u5931\uff08Lbd\uff09\u4e0e \u6548\u7528\u635f\u5931\uff08Lcl\uff09<\/strong>\u7684\u8054\u5408\u4f18\u5316\u6846\u67b6\uff0c\u5e73\u8861\u653b\u51fb\u6210\u529f\u7387\u4e0e\u6a21\u578b\u6027\u80fd\u4fdd\u6301<\/li>\n\n\n\n<li>\u5f15\u5165<strong>\u5d4c\u5165\u635f\u5931\uff08Lemb\uff09<\/strong>\u7ea6\u675f\u89c6\u89c9\u7f16\u7801\u5668\uff0c\u901a\u8fc7\u6559\u5e08\u6a21\u578b\u84b8\u998f\u4fdd\u6301\u591a\u6a21\u6001\u5bf9\u9f50\u7279\u6027<\/li>\n<\/ul>\n\n\n\n<p><strong>3. \u8de8\u4efb\u52a1\u6cdb\u5316\u6027<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5728\u56fe\u50cf\u63cf\u8ff0\u3001\u89c6\u89c9\u95ee\u7b54\u7b49\u4efb\u52a1\u4e2d\u9a8c\u8bc1\u653b\u51fb\u6709\u6548\u6027\uff08ASR>95%\uff09\uff0c\u4e14\u80fd\u89c4\u907f\u73b0\u6709\u9632\u5fa1\uff08\u5982\u8f93\u5165\u51c0\u5316\u3001\u5fae\u8c03\uff09<\/li>\n\n\n\n<li>\u5728\u81ea\u52a8\u9a7e\u9a76\uff08\u4ea4\u901a\u706f\u72b6\u6001\u7be1\u6539\uff09\u548c\u533b\u7597\u8bca\u65ad\uff08\u6076\u610f\u7528\u836f\u5efa\u8bae\u6ce8\u5165\uff09\u573a\u666f\u4e2d\u9a8c\u8bc1\u5b9e\u9645\u5a01\u80c1<\/li>\n<\/ul>\n\n\n\n<p><strong>4. \u9690\u853d\u6027\u589e\u5f3a\u673a\u5236<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u901a\u8fc7\u89e6\u53d1\u5668\u4e0e\u76ee\u6807\u8bcd\u5143\u7684\u7ec6\u7c92\u5ea6\u5173\u8054\uff0c\u5b9e\u73b0\u4ec5\u4fee\u6539\u5173\u952e\u8bcd\u5143 \u7684\u9ad8\u9690\u853d\u653b\u51fb\uff08ATS>0.75\uff09<\/li>\n\n\n\n<li>\u652f\u6301\u591a\u76ee\u6807\u653b\u51fb\uff08\u540c\u65f6\u5d4c\u5165\u591a\u4e2a\u89e6\u53d1\u5668-\u76ee\u6807\u8bcd\u5143\u5bf9\uff09<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"h_32273152617_2\"><a href=\"https:\/\/zhida.zhihu.com\/search?content_id=255476962&amp;content_type=Article&amp;match_order=1&amp;q=AutoDrive-QA&amp;zhida_source=entity\" target=\"_blank\" rel=\"noreferrer noopener\">AutoDrive-QA<\/a><\/h2>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"h_32273152617_2\">&#8211; Automated Generation<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8bba\u6587\u6807\u9898\uff1aAutoDrive-QA- Automated Generation of Multiple-Choice Questions for Autonomous Driving Datasets Using Large Vision-Language Models<\/li>\n\n\n\n<li>\u8bba\u6587\u94fe\u63a5\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/arxiv.org\/abs\/2503.15778\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/arxiv.org\/abs\/2503.15778<\/a><\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/pica.zhimg.com\/v2-45fd796c187c4e674ce9c54084bc57d6_1440w.jpg\" alt=\"\"\/><\/figure>\n\n\n\n<p><strong>\u6838\u5fc3\u521b\u65b0\u70b9\uff1a<\/strong><\/p>\n\n\n\n<p><strong>1. \u7edf\u4e00\u7ed3\u6784\u5316\u8bc4\u4f30\u6846\u67b6\u7684\u6784\u5efa<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u63d0\u51faAutoDrive-QA\u57fa\u51c6\u6d4b\u8bd5\uff0c\u901a\u8fc7\u6574\u5408\u591a\u6e90\u5f02\u6784\u6570\u636e\u96c6\uff08DriveLM\u3001NuScenes-QA\u3001LingoQA\uff09\uff0c\u9996\u6b21\u7cfb\u7edf\u8986\u76d6\u81ea\u52a8\u9a7e\u9a76\u4e09\u5927\u6838\u5fc3\u4efb\u52a1\u2014\u2014\u611f\u77e5\uff08Perception\uff09\u3001\u9884\u6d4b\uff08Prediction\uff09\u548c\u89c4\u5212\uff08Planning\uff09\uff0c\u5e76\u6807\u51c6\u5316\u95ee\u7b54\u683c\u5f0f\uff0c\u89e3\u51b3\u4e86\u73b0\u6709\u8bc4\u4f30\u4f53\u7cfb\u788e\u7247\u5316\u7684\u95ee\u9898\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>2. \u9886\u57df\u5bfc\u5411\u7684\u5e72\u6270\u9879\u81ea\u52a8\u751f\u6210\u6280\u672f<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u57fa\u4e8e\u5927\u578b\u89c6\u89c9\u8bed\u8a00\u6a21\u578b\uff08VLMs\uff09\u8bbe\u8ba1\u81ea\u52a8\u5316\u6d41\u6c34\u7ebf\uff0c\u7ed3\u5408\u9a7e\u9a76\u9886\u57df\u9519\u8bef\u6a21\u5f0f\uff08\u5982\u4f20\u611f\u5668\u8bef\u5224\u3001\u903b\u8f91\u4e0d\u4e00\u81f4\u3001\u9886\u57df\u6982\u5ff5\u8bef\u89e3\u7b49\uff09\uff0c\u751f\u6210\u9ad8\u771f\u5b9e\u6027\u5e72\u6270\u9879\uff08Distractors\uff09\u3002\u4e0e\u4f20\u7edf\u65b9\u6cd5\u76f8\u6bd4\uff0c\u8be5\u65b9\u6cd5\u901a\u8fc7\u4efb\u52a1\u7279\u5b9a\u9519\u8bef\u5efa\u6a21\uff08Task-Specific Error Modeling\uff09\u663e\u8457\u63d0\u5347\u5e72\u6270\u9879\u7684\u8bed\u4e49\u76f8\u5173\u6027\u4e0e\u6311\u6218\u6027\uff0c\u51cf\u5c11\u56e0\u5e72\u6270\u9879\u8fc7\u4e8e\u7b80\u5355\u5bfc\u81f4\u7684\u8bc4\u4f30\u504f\u5dee\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>3. \u96f6\u6837\u672c\u6027\u80fd\u8bc4\u4f30\u4e0e\u7ec6\u7c92\u5ea6\u9519\u8bef\u5206\u6790<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5728\u96f6\u6837\u672c\uff08Zero-Shot\uff09\u8bbe\u7f6e\u4e0b\uff0c\u9a8c\u8bc1\u591a\u6a21\u578b\uff08\u5982GPT-4V\u3001Qwen2-VL\u7cfb\u5217\uff09\u7684\u6027\u80fd\u8fb9\u754c\u3002\u5b9e\u9a8c\u8868\u660e\uff0c\u6a21\u578b\u5728\u611f\u77e5\u4efb\u52a1\uff08\u5e73\u574774.94%\uff09\u8868\u73b0\u6700\u4f18\uff0c\u9884\u6d4b\u4efb\u52a1\uff0865.33%\uff09\u6311\u6218\u6700\u5927\u3002\u8fdb\u4e00\u6b65\u901a\u8fc7\u5e72\u6270\u9879\u9009\u62e9\u8ffd\u8e2a\uff0c\u91cf\u5316\u9519\u8bef\u7c7b\u578b\u5206\u5e03\uff08\u5982\u611f\u77e5\u4efb\u52a1\u4e2d45.74%\u4e3a\u9886\u57df\u6982\u5ff5\u9519\u8bef\uff09\uff0c\u63ed\u793a\u6a21\u578b\u5728\u89c6\u89c9\u7406\u89e3\u3001\u903b\u8f91\u63a8\u7406\u7b49\u7ef4\u5ea6\u7684\u8584\u5f31\u73af\u8282\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>4. \u53ef\u6269\u5c55\u7684\u8bc4\u4f30\u65b9\u6cd5\u8bba\u521b\u65b0<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u63d0\u51fa\u591a\u667a\u80fd\u4f53\u534f\u540c\u8fc7\u6ee4\u673a\u5236\uff0c\u7ed3\u5408\u751f\u6210-\u8bc4\u4f30-\u4f18\u5316\u5faa\u73af\uff0c\u786e\u4fdd\u751f\u6210\u7684\u591a\u9009\u9898\uff08MCQ\uff09\u6ee1\u8db3\u5355\u6b63\u786e\u89e3\u3001\u65e0\u5197\u4f59\u5e72\u6270\u9879\u7b49\u7ea6\u675f\u6761\u4ef6\uff0c\u540c\u65f6\u652f\u6301\u5927\u89c4\u6a21\u6570\u636e\u96c6\u7684\u9ad8\u6548\u8f6c\u6362\uff08\u5982\u5904\u7406\u8d8550\u4e07QA\u5bf9\uff09\uff0c\u4e3a\u81ea\u52a8\u9a7e\u9a76\u6a21\u578b\u7684\u8fed\u4ee3\u4e0e\u8de8\u6570\u636e\u96c6\u6cdb\u5316\u63d0\u4f9b\u6807\u51c6\u5316\u5de5\u5177\u3002<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"h_32273152617_3\"><a href=\"https:\/\/zhida.zhihu.com\/search?content_id=255476962&amp;content_type=Article&amp;match_order=1&amp;q=GASP&amp;zhida_source=entity\" target=\"_blank\" rel=\"noreferrer noopener\">GASP<\/a><\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u8bba\u6587\u6807\u9898\uff1aGASP: Unifying Geometric and Semantic Self-Supervised Pre-training for Autonomous Driving<\/li>\n\n\n\n<li>\u8bba\u6587\u94fe\u63a5\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/arxiv.org\/abs\/2503.15672\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/arxiv.org\/abs\/2503.15672<\/a><\/li>\n\n\n\n<li>\u9879\u76ee\u4e3b\u9875\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/research.zenseact.com\/publications\/gasp\/\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/research.zenseact.com\/publications\/gasp\/<\/a><\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image\"><img decoding=\"async\" src=\"https:\/\/pica.zhimg.com\/v2-12acf1cd2c0544bb4c556a5f6401406e_1440w.jpg\" alt=\"\"\/><\/figure>\n\n\n\n<p><strong>\u6838\u5fc3\u521b\u65b0\u70b9\uff1a<\/strong><\/p>\n\n\n\n<p><strong>1. \u51e0\u4f55-\u8bed\u4e49\u8054\u5408\u9884\u8bad\u7ec3\u6846\u67b6<\/strong><\/p>\n\n\n\n<p>\u63d0\u51fa\u7edf\u4e00\u76844D\u65f6\u7a7a\u8868\u5f81\u5b66\u4e60\u8303\u5f0f \uff0c\u901a\u8fc7\u4e09\u4e2a\u534f\u540c\u4efb\u52a1\u5b9e\u73b0\u591a\u6a21\u6001\u7279\u5f81\u878d\u5408\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u51e0\u4f55\u5360\u7528\u9884\u6d4b \uff1a\u5efa\u6a21\u52a8\u6001\u573a\u666f\u7684\u8fde\u7eed4D\u5360\u7528\u573a\uff083D\u7a7a\u95f4+\u65f6\u95f4\uff09\uff0c\u6355\u6349\u73af\u5883\u7ed3\u6784\u6f14\u53d8<\/li>\n\n\n\n<li>\u8bed\u4e49\u7279\u5f81\u84b8\u998f \uff1a\u9884\u6d4b<a href=\"https:\/\/zhida.zhihu.com\/search?content_id=255476962&amp;content_type=Article&amp;match_order=1&amp;q=DINOv2&amp;zhida_source=entity\" target=\"_blank\" rel=\"noreferrer noopener\">DINOv2<\/a><\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u89c6\u89c9\u57fa\u7840\u6a21\u578b\u7684\u9ad8\u5c42\u8bed\u4e49\u7279\u5f81\uff0c\u6ce8\u5165\u8bed\u4e49\u5148\u9a8c<\/li>\n\n\n\n<li>\u81ea\u8f66\u8def\u5f84\u9884\u6d4b \uff1a\u901a\u8fc7\u9690\u5f0f ego-path occupancy \u5efa\u6a21\u8f66\u8f86\u53ef\u884c\u7a7a\u95f4\uff0c\u589e\u5f3a\u8fd0\u52a8\u7406\u89e3<\/li>\n<\/ul>\n\n\n\n<p><strong>2. \u65f6\u7a7a\u8fde\u7eed\u573a\u8868\u793a<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u7a81\u7834\u4f20\u7edf\u79bb\u6563\u5316\u9884\u6d4b\u5c40\u9650\uff0c\u91c7\u7528\u9690\u5f0f\u795e\u7ecf\u573a\uff08Implicit Neural Fields\uff09 \u8868\u5f81\u65f6\u7a7a\u8fde\u7eed\u7684\u51e0\u4f55-\u8bed\u4e49\u4fe1\u606f\uff0c\u652f\u6301\u4efb\u610f\u65f6\u7a7a\u5750\u6807\uff08x,y,z,t\uff09\u7684\u67e5\u8be2\uff0c\u63d0\u5347\u52a8\u6001\u573a\u666f\u6cdb\u5316\u80fd\u529b\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>3. \u591a\u6e90\u81ea\u76d1\u7763\u4fe1\u53f7\u878d\u5408<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u521b\u65b0\u6027\u6574\u5408\u4e09\u7c7b\u81ea\u76d1\u7763\u4fe1\u53f7\uff1a<\/li>\n\n\n\n<li>LiDAR\u5c04\u7ebf\u51e0\u4f55\u7ea6\u675f \uff08\u542b\u7f3a\u5931\u5c04\u7ebf\u8d1f\u6837\u672c\u6316\u6398\uff09<\/li>\n\n\n\n<li>\u56fe\u50cf\u8bed\u4e49\u7279\u5f81\u84b8\u998f \uff08\u57fa\u4e8e\u4f4d\u7f6e\u7f16\u7801\u53bb\u566a\u7684DINOv2\u7279\u5f81\uff09<\/li>\n\n\n\n<li>\u81ea\u8f66\u8fd0\u52a8\u8f68\u8ff9\u5148\u9a8c \uff08\u591a\u6a21\u6001\u8def\u5f84\u6982\u7387\u573a\u5efa\u6a21\uff09<\/li>\n<\/ul>\n\n\n\n<p><strong>4. \u6570\u636e\u589e\u5f3a\u7b56\u7565<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u63d0\u51fa\u65cb\u8f6c\u589e\u5f3a\uff08\u00b120\u00b0\u968f\u673a\u65cb\u8f6c\uff09 \u6d88\u9664\u65b9\u5411\u504f\u7f6e\uff0c\u7ed3\u5408\u7f3a\u5931LiDAR\u5c04\u7ebf\u63a8\u7406 \u63d0\u5347\u51e0\u4f55\u8868\u5f81\u9c81\u68d2\u6027\uff0c\u5728<a href=\"https:\/\/zhida.zhihu.com\/search?content_id=255476962&amp;content_type=Article&amp;match_order=1&amp;q=Argoverse2&amp;zhida_source=entity\" target=\"_blank\" rel=\"noreferrer noopener\">Argoverse2<\/a><\/li>\n<\/ul>\n\n\n\n<p>\u7b49\u6570\u636e\u96c6\u5b9e\u73b0\u8bed\u4e49\u5360\u7528\u9884\u6d4b\uff08\u219115.5mAP\uff09\u3001\u5728\u7ebf\u5efa\u56fe\uff08\u21915.8mIoU\uff09\u7b49\u4e0b\u6e38\u4efb\u52a1\u7684\u663e\u8457\u63d0\u5347\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u539f\u6587\u94fe\u63a5\uff1ahttps:\/\/zhuanlan.z [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3,1],"tags":[],"class_list":["post-26632","post","type-post","status-publish","format-standard","hentry","category-technology-frontier","category-home"],"views":0,"_links":{"self":[{"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/posts\/26632"}],"collection":[{"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/comments?post=26632"}],"version-history":[{"count":1,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/posts\/26632\/revisions"}],"predecessor-version":[{"id":26633,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/posts\/26632\/revisions\/26633"}],"wp:attachment":[{"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/media?parent=26632"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/categories?post=26632"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/222.128.65.89:18086\/index.php\/wp-json\/wp\/v2\/tags?post=26632"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}