{"id":10449,"date":"2026-07-26T17:34:09","date_gmt":"2026-07-26T09:34:09","guid":{"rendered":"https:\/\/infernews.com\/blog\/provise\/"},"modified":"2026-07-26T17:36:31","modified_gmt":"2026-07-26T09:36:31","slug":"provise","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/provise\/","title":{"rendered":"ProVisE \u7528\u50cf\u7d20\u7b54\u6848\u91cd\u505a\u7a7a\u9593\u8a55\u6e2c"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/07\/pasted-b002ac0fb047.jpg\" alt=\"ProVisE logo\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u7576\u4e00\u689d\u7a7a\u9593\u984c\u76ee\u672c\u4f86\u61c9\u8a72\u7528\u5708\u9078\u3001\u6a19\u8a18\u8def\u5f91\u6216\u8005\u906e\u7f69\u53bb\u8868\u9054\uff0c\u786c\u8981\u6a21\u578b\u4ea4\u51fa\u5ea7\u6a19\u3001\u9078\u9805\u5b57\u6bcd\u6216\u6587\u5b57\u63cf\u8ff0\uff0c\u7d50\u679c\u5f80\u5f80\u5514\u4fc2\u80fd\u529b\u5dee\uff0c\u800c\u4fc2\u7b54\u984c\u4ecb\u9762\u540c\u6a21\u578b\u8868\u9054\u65b9\u5f0f\u932f\u4f4d\u3002ProVisE\u5c6c\u65bc\u8a55\u6e2c\u6846\u67b6\uff0c\u8655\u7406\u7684\u6b63\u662f\u5462\u500b\u843d\u5dee\uff1a\u5b83\u5514\u6539\u539f\u672c benchmark \u4efb\u52d9\u672c\u8eab\uff0c\u53ea\u6539\u56de\u61c9\u4ecb\u9762\uff0c\u8b93\u5716\u50cf\u751f\u6210\u6a21\u578b\u7528\u50cf\u7d20\u7a7a\u9593\u4ea4\u7b54\u6848\uff0c\u518d\u8f49\u56de benchmark \u53ef\u8a08\u5206\u7684\u7d50\u69cb\u5316\u8f38\u51fa\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u73fe\u6709 spatial benchmarks \u591a\u6578\u6cbf\u7528 text-only interface\uff0c\u5047\u8a2d\u6240\u6709\u6a21\u578b\u90fd\u61c9\u8a72\u4ee5 coordinates\u3001option labels \u6216 textual descriptions \u56de\u7b54\u3002\u4f5c\u8005\u8a8d\u70ba\u9019\u7a2e\u56fa\u5b9a\u7bc4\u5f0f\u6703\u58d3\u7e2e regions\u3001paths\u3001affordances \u5462\u985e\u672c\u8eab\u504f\u8996\u89ba\u7684\u5224\u65b7\uff0c\u56e0\u6b64\u63d0\u51fa Protocolized Visual Evaluation\uff1a\u5148\u7531 task-aware router \u6307\u6d3e visual protocol\uff0c\u518d\u7528\u56fa\u5b9a guidance prompt \u540c parser \u7d04\u675f\u8f38\u51fa\uff0c\u6700\u5f8c\u4ecd\u7136\u4ea4\u56de original benchmark metric \u8a55\u5206\u3002Text-output VLMs \u5c31\u7dad\u6301\u539f\u672c\u7b54\u984c\u7a7a\u9593\uff0c\u5169\u985e\u6a21\u578b\u53ef\u4ee5\u5728\u540c\u4e00\u5957\u4efb\u52d9\u8a9e\u7fa9\u4e0b\u6bd4\u8f03\u3002<\/p>\n\n\n<div\n    \tclass=\"align wp-block-vpb-video\"    id='vpbpVideoPlayer-1'\n    data-attributes='{&quot;source&quot;:&quot;https:\\\/\\\/zju-omniai.github.io\\\/ProVisE\\\/assets\\\/provise-overview.mp4&quot;,&quot;repeat&quot;:true,&quot;autoplay&quot;:true,&quot;muted&quot;:true,&quot;align&quot;:&quot;&quot;,&quot;poster&quot;:&quot;&quot;,&quot;controls&quot;:{&quot;play-large&quot;:true,&quot;restart&quot;:false,&quot;rewind&quot;:true,&quot;play&quot;:true,&quot;fast-forward&quot;:true,&quot;progress&quot;:true,&quot;current-time&quot;:true,&quot;duration&quot;:false,&quot;mute&quot;:true,&quot;volume&quot;:true,&quot;pip&quot;:false,&quot;airplay&quot;:false,&quot;captions&quot;:false,&quot;settings&quot;:true,&quot;download&quot;:false,&quot;fullscreen&quot;:true},&quot;captions&quot;:[],&quot;ratio&quot;:&quot;16:9&quot;,&quot;accentColor&quot;:&quot;&quot;,&quot;lazyLoad&quot;:false,&quot;width&quot;:&quot;100%&quot;,&quot;radius&quot;:&quot;0px&quot;,&quot;resetOnEnd&quot;:false,&quot;autoHideControl&quot;:true,&quot;isSetup&quot;:false}'\n><\/div>\n\n\n<p class=\"wp-block-paragraph\">ZJU-OmniAI\/ProVisE \u5728\u65bc\u628a\u300c\u6a21\u578b\u5514\u8b58\u7b54\u300d\u540c\u300c\u8a55\u6e2c\u65b9\u6cd5\u903c\u932f\u7b54\u6848\u683c\u5f0f\u300d\u5206\u958b\u8655\u7406\u3002\u914d\u5957\u7684 SpatialGen-Bench \u6536\u9304 470 \u500b curated samples\uff0c\u6db5\u84cb 14 \u500b subtasks\uff0c\u540c\u6642\u5206\u6210 perception\u3001understanding\u3001reasoning\u3001interaction \u56db\u500b capability levels\uff1b\u7814\u7a76\u7d50\u8ad6\u4ea6\u76f8\u7576\u76f4\u63a5\uff0cimage-generation models \u5728\u53ef\u628a\u5224\u65b7\u5916\u5316\u6210\u50cf\u7d20\u6a19\u8a18\u7684\u4efb\u52d9\u4e0a\u6709\u7af6\u722d\u529b\uff0c\u4f46 text-output VLMs \u5728\u53e6\u5916\u4e00\u4e9b\u984c\u578b\u4ecd\u7136\u8f03\u7a69\u5b9a\uff0c\u5169\u8005\u4e26\u975e\u8ab0\u5168\u9762\u53d6\u4ee3\u8ab0\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4fdd\u7559\u539f\u6709 benchmark metric\uff0c\u53ea\u66ff\u63db\u7b54\u6848\u4ecb\u9762\uff0c\u65b9\u4fbf\u540c\u65e2\u6709\u7d50\u679c\u5c0d\u7167<\/li>\n\n\n\n<li>\u7528 visual protocol \u9650\u5236\u751f\u6210\u5167\u5bb9\uff0c\u6e1b\u5c11\u4efb\u610f\u756b\u5716\u5e36\u4f86\u7684\u89e3\u6790\u6b67\u7fa9<\/li>\n\n\n\n<li>SpatialGen-Bench \u628a\u7a7a\u9593\u80fd\u529b\u62c6\u6210 14 \u500b subtasks\uff0c\u5514\u518d\u53ea\u770b\u55ae\u4e00\u7e3d\u5206<\/li>\n\n\n\n<li>\u9069\u5408\u7814\u7a76 VLM\u3001image-generation models\u3001agent \u7a7a\u9593\u7406\u89e3\u80fd\u529b\u7684\u5718\u968a\u63a1\u7528<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b89\u88dd\u9580\u6abb\u770b\u4f86\u4e0d\u9ad8\uff0c\u7a0b\u5f0f\u74b0\u5883\u4ee5 Python 3.10+ \u70ba\u4e3b\uff0c\u4e26\u5df2\u516c\u958b code\u3001project page \u8207 Hugging Face \u4e0a\u7684 SpatialGen-Bench\u3002\u73fe\u968e\u6bb5\u5b83\u66f4\u50cf\u7814\u7a76\u8207\u8a55\u6e2c\u9805\u76ee\uff0c\u4e0d\u662f\u5373\u63d2\u5373\u7528\u7522\u54c1\uff1b\u91cd\u9ede\u4e5f\u4e0d\u5728\u90e8\u7f72\u6210\u670d\u52d9\uff0c\u800c\u662f\u5728\u4f60\u60f3\u9a57\u8b49\u6a21\u578b\u7a7a\u9593\u8a8d\u77e5\u6642\uff0c\u80fd\u5426\u7528\u66f4\u8cbc\u8fd1\u6a21\u578b\u8f38\u51fa\u5f62\u5f0f\u7684\u65b9\u5f0f\u505a\u6bd4\u8f03\u3002\u5c0d\u505a\u591a\u6a21\u614b\u6a21\u578b\u3001\u8996\u89ba\u8a55\u6e2c\u6216 Agentic \u7cfb\u7d71\u7684\u4eba\u4f86\u8aaa\uff0cProVisE\u63d0\u4f9b\u4e86\u4e00\u500b\u76f8\u7576\u6e05\u6670\u7684\u6aa2\u67e5\u89d2\u5ea6\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/zju-omniai.github.io\/ProVisE\/\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/ZJU-OmniAI\/ProVisE\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a> \u00b7 <a href=\"https:\/\/arxiv.org\/pdf\/2607.21072\" rel=\"noopener noreferrer\"><strong>Paper<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u7a7a\u9593\u7406\u89e3\u5514\u4e00\u5b9a\u9069\u5408\u7528\u6587\u5b57\u4f5c\u7b54\uff0cProVisE\u6539\u5497\u8a55\u6e2c\u5165\u53e3\uff0c\u4ee4\u5716\u50cf\u751f\u6210\u6a21\u578b\u53ef\u4ee5\u7528\u756b\u3001\u9ede\u3001\u6a19\u8a18\u53bb\u7b54\u984c\u3002\u4f60\u6703\u66f4\u6613\u7747\u6e05\u695a\uff0c\u6a21\u578b\u5230\u5e95\u4fc2\u5514\u8b58\u7a7a\u9593\uff0c\u5b9a\u53ea\u4fc2\u7b54\u984c\u683c\u5f0f\u5514\u5c0d\u3002<\/p>\n","protected":false},"author":8,"featured_media":10448,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,116,30,152,119,149,199],"tags":[],"class_list":["post-10449","post","type-post","status-publish","format-standard","hentry","category-133","category-agentic","category-image","category-python","category-119","category-149","category-dataset-"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10449","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=10449"}],"version-history":[{"count":1,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10449\/revisions"}],"predecessor-version":[{"id":10451,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10449\/revisions\/10451"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/10448"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=10449"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=10449"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=10449"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}