{"id":8639,"date":"2026-05-30T19:44:11","date_gmt":"2026-05-30T11:44:11","guid":{"rendered":"https:\/\/infernews.com\/blog\/towards-consistent-video-geometry-estimation\/"},"modified":"2026-05-30T19:48:57","modified_gmt":"2026-05-30T11:48:57","slug":"towards-consistent-video-geometry-estimation","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/towards-consistent-video-geometry-estimation\/","title":{"rendered":"ViGeo\uff1a\u4e00\u500b\u6a21\u578b\u8655\u7406\u5f71\u7247\u5e7e\u4f55\u91cd\u5efa"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/05\/pasted-3633ee9721c3.jpg\" alt=\"Repository image for aigc3d\/ViGeo\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">ViGeo \u662f\u4e00\u500b\u7528\u4f86\u4f30\u7b97\u5834\u666f\u5e7e\u4f55\u7684\u9805\u76ee\uff0c\u8f38\u5165\u53ef\u4ee5\u662f\u5f71\u7247\u7247\u6bb5\uff0c\u4e5f\u53ef\u4ee5\u662f\u55ae\u5f35\u5f71\u50cf\u3002\u5b83\u6703\u8f38\u51fa depth\u30013D points\u3001normals\u3001confidence\uff0c\u8655\u7406\u9023\u7e8c\u5f71\u683c\u6642\u4ea6\u53ef\u4f30\u7b97 camera poses\uff0c\u91cd\u9ede\u662f\u76e1\u91cf\u4fdd\u6301\u6642\u9593\u4e0a\u7684\u4e00\u81f4\u6027\uff0c\u6e1b\u5c11\u524d\u5f8c\u5e40\u7d50\u679c\u8df3\u52d5\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u4f7f\u7528\u9019\u500b\u9805\u76ee\u6642\uff0c\u5148\u6309\u624b\u982d\u8cc7\u6599\u9078\u64c7\u6a21\u5f0f\uff1a\u5b8c\u6574\u5f71\u7247\u53ef\u7528 offline\uff0c\u4e32\u6d41\u756b\u9762\u53ef\u7528 online\uff0c\u9577\u5f71\u7247\u5247\u53ef\u5206\u6bb5\u7528 chunk \u8655\u7406\u3002\u9019\u7a2e\u5b89\u6392\u5c0d\u505a\u5f71\u7247\u91cd\u5efa\u3001\u6a5f\u68b0\u4eba\u611f\u77e5\u3001AR\u3001\u5c0e\u822a\u6216\u5f8c\u671f\u8996\u89ba\u5206\u6790\u7684\u4eba\u8f03\u5be6\u7528\uff0c\u56e0\u70ba\u4e0d\u9700\u8981\u70ba\u4e0d\u540c\u8f38\u5165\u5f62\u5f0f\u63db\u53e6\u4e00\u5957\u6a21\u578b\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b83\u60f3\u89e3\u6c7a\u7684\u6838\u5fc3\u554f\u984c\uff0c\u662f\u5f71\u7247\u5e7e\u4f55\u4f30\u8a08\u5e38\u898b\u7684\u5169\u96e3\uff1a\u4e0d\u662f\u77ed\u7247\u6548\u679c\u597d\u4f46\u96e3\u4ee5\u4e32\u6d41\uff0c\u5c31\u662f\u80fd\u5373\u6642\u63a8\u7406\u4f46\u9577\u6642\u9593\u4e00\u81f4\u6027\u4e0d\u8db3\u3002ViGeo \u4ee5\u540c\u4e00\u500b feed-forward foundation model \u7d71\u4e00 full-sequence reconstruction\u3001streaming inference \u8207 long-video inference\uff0c\u8ad6\u6587\u6307\u51fa\u95dc\u9375\u5728 dynamic chunking attention\uff0c\u8b93\u6a21\u578b\u53ef\u56e0\u61c9\u6e2c\u8a66\u60c5\u5883\u5207\u63db\u6642\u9593\u95dc\u6ce8\u65b9\u5f0f\uff0c\u800c\u4e0d\u7528\u91cd\u65b0\u8a13\u7df4\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u53e6\u4e00\u500b\u91cd\u8981\u90e8\u5206\u662f VideoLDCM\uff0c\u5b8c\u6574\u540d\u7a31\u662f VideoLDCM\uff0c\u8ca0\u8cac depth completion\u3002\u5b83\u5728\u9019\u9805\u5de5\u4f5c\u4e2d\u7528\u4f5c data-refinement model\uff0c\u628a\u7a00\u758f\u6216\u5e36\u96dc\u8a0a\u7684\u6df1\u5ea6\u89c0\u6e2c\u6574\u7406\u6210\u8f03\u4e7e\u6de8\u7684 dense depth supervision\uff0c\u5c0d\u8a13\u7df4\u5e7e\u4f55\u6a21\u578b\u6709\u5e6b\u52a9\uff0c\u4e5f\u89e3\u91cb\u4e86\u70ba\u4f55\u9019\u500b\u9805\u76ee\u4e0d\u53ea\u770b\u55ae\u5e40\u54c1\u8cea\uff0c\u9084\u5f37\u8abf\u8de8\u5f71\u683c\u7a69\u5b9a\u6027\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u540c\u6642\u652f\u63f4 offline\u3001online\u3001chunk \u4e09\u7a2e\u63a8\u7406\u6d41\u7a0b<\/li>\n\n\n\n<li>\u53ef\u7531\u5f71\u7247\u6216\u55ae\u5f35\u5f71\u50cf\u4f30\u7b97 depth\u30013D points\u3001normals \u7b49\u7d50\u679c<\/li>\n\n\n\n<li>\u4ee5 dynamic chunking attention \u517c\u9867\u4e32\u6d41\u8207\u9577\u5f71\u7247\u8655\u7406<\/li>\n\n\n\n<li>\u7d50\u5408 VideoLDCM \u6539\u5584\u6df1\u5ea6\u76e3\u7763\u8cc7\u6599\u54c1\u8cea<\/li>\n\n\n\n<li>\u8ad6\u6587\u8072\u7a31\u5728\u591a\u9805 video geometry \u4efb\u52d9\u9054\u5230 state-of-the-art<\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th>Model<\/th><th>Download<\/th><th>Description<\/th><\/tr><\/thead><tbody><tr><td>ViGeo<\/td><td><a href=\"https:\/\/huggingface.co\/pkqbajng\/ViGeo\">LINK<\/a><\/td><td>\u7528\u65bc\u6df1\u5ea6\u3001\u9ede\u3001\u6cd5\u7dda\u3001\u59ff\u614b\u548c\u7f6e\u4fe1\u5ea6\u7684\u4e3b\u8981\u8996\u89ba\u5e7e\u4f55\u6a21\u578b<\/td><\/tr><tr><td>VideoLDCM<\/td><td><a href=\"https:\/\/huggingface.co\/pkqbajng\/VideoLDCM\">LINK<\/a><\/td><td>\u7528\u65bc\u7a00\u758f\u6df1\u5ea6\u6ffe\u6ce2\u3001\u6cca\u677e\u88dc\u5168\u548c\u6df1\u5ea6\u7d30\u5316\u7684\u8cc7\u6599\u7d30\u5316\u6a21\u578b<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u6027\u80fd\u65b9\u9762\uff0c\u8ad6\u6587\u63cf\u8ff0\u5b83\u5728 online\u3001offline\u3001long-video depth estimation\u3001surface normal estimation\u3001video point map estimation \u90fd\u6709\u5f88\u5f37\u8868\u73fe\uff0c\u4e26\u4ee5 public datasets \u8a13\u7df4\u3002\u4e0d\u904e\u76ee\u524d\u516c\u958b checkpoint \u4ea6\u5df2\u8a3b\u660e\u5b58\u5728\u5df2\u77e5 loss implementation \u554f\u984c\uff0c\u53ef\u80fd\u5728 camera poses \u8996\u89ba\u5316\u8207\u9060\u8ddd\u5340\u57df\u51fa\u73fe\u8f15\u5fae\u7455\u75b5\uff0c\u56e0\u6b64\u8f03\u9069\u5408\u5148\u7528\u4f86\u7406\u89e3\u80fd\u529b\u7bc4\u570d\uff0c\u518d\u6c7a\u5b9a\u662f\u5426\u653e\u5165\u8981\u6c42\u5f88\u9ad8\u7684\u751f\u7522\u6d41\u7a0b\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>GitHub\uff1a<\/strong> <a href=\"https:\/\/github.com\/aigc3d\/ViGeo\" rel=\"noopener noreferrer\">https:\/\/github.com\/aigc3d\/ViGeo<\/a><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>\u9805\u76ee\uff1a<\/strong> <a href=\"https:\/\/pkqbajng.github.io\/ViGeo\/\" rel=\"noopener noreferrer\">https:\/\/pkqbajng.github.io\/ViGeo\/<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>ViGeo\u628a\u5f71\u7247\u6216\u55ae\u5f35\u756b\u9762\u7684\u5e7e\u4f55\u8cc7\u8a0a\u4e00\u6b21\u4f30\u51fa\u4f86\u3002\u5c0d\u6df1\u5ea6\u3001\u6cd5\u7dda\u30013D\u9ede\u8207\u93e1\u982d\u4f4d\u59ff\u6709\u8208\u8da3\u7684\u4eba\uff0c\u9019\u500b\u9805\u76ee\u5f88\u503c\u5f97\u7559\u610f\u3002<\/p>\n","protected":false},"author":8,"featured_media":8638,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","wpai_meta_description":"","footnotes":""},"categories":[133,171,76,149,186],"tags":[],"class_list":["post-8639","post","type-post","status-publish","format-standard","hentry","category-133","category-171","category-76","category-149","category-186"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/8639","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/comments?post=8639"}],"version-history":[{"count":1,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/8639\/revisions"}],"predecessor-version":[{"id":8642,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/8639\/revisions\/8642"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/8638"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=8639"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=8639"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=8639"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}