{"id":34453,"date":"2024-12-06T22:56:23","date_gmt":"2024-12-06T14:56:23","guid":{"rendered":"https:\/\/17aitech.com\/?p=34453"},"modified":"2024-12-06T22:56:23","modified_gmt":"2024-12-06T14:56:23","slug":"%e5%bc%80%e6%ba%90%e8%a7%86%e9%a2%91%e7%89%88gpt-4o%ef%bc%9f%e5%bf%ab%e9%80%9f%e8%ae%b0%e5%bf%86%ef%bc%8c%e5%ae%9e%e6%97%b6%e9%97%ae%e7%ad%94%ef%bc%8c%e6%8b%bf%e4%b8%8bcvpr24%e9%95%bf%e8%a7%86","status":"publish","type":"post","link":"https:\/\/17aitech.com\/?p=34453","title":{"rendered":"\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR&#8217;24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b"},"content":{"rendered":"<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.jiqizhixin.com\/articles\/2024-07-08-12\" target=\"_blank\">\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR&#8217;24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b<\/a><\/p>\n<section><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-ace45b8b0af65f751c1a14194401fe7b.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-ace45b8b0af65f751c1a14194401fe7b.png\"><\/a><\/section>\n<blockquote data-author-name=\"\" data-content-utf8-length=\"166\" data-source-title=\"\" data-type=\"2\" data-url=\"\">\n<section>\n<section>AIxiv\u4e13\u680f\u662f<mark data-type=\"institutions\" data-id=\"9e96434f-5827-41e3-9b3c-36d39bc0d446\">\u673a\u5668\u4e4b\u5fc3<\/mark>\u53d1\u5e03\u5b66\u672f\u3001\u6280\u672f\u5185\u5bb9\u7684\u680f\u76ee\u3002\u8fc7\u53bb\u6570\u5e74\uff0c<mark data-type=\"institutions\" data-id=\"9e96434f-5827-41e3-9b3c-36d39bc0d446\">\u673a\u5668\u4e4b\u5fc3<\/mark>AIxiv\u4e13\u680f\u63a5\u6536\u62a5\u9053\u4e862000\u591a\u7bc7\u5185\u5bb9\uff0c\u8986\u76d6\u5168\u7403\u5404\u5927\u9ad8\u6821\u4e0e\u4f01\u4e1a\u7684\u9876\u7ea7\u5b9e\u9a8c\u5ba4\uff0c\u6709\u6548\u4fc3\u8fdb\u4e86\u5b66\u672f\u4ea4\u6d41\u4e0e\u4f20\u64ad\u3002\u5982\u679c\u60a8\u6709\u4f18\u79c0\u7684\u5de5\u4f5c\u60f3\u8981\u5206\u4eab\uff0c\u6b22\u8fce\u6295\u7a3f\u6216\u8005\u8054\u7cfb\u62a5\u9053\u3002\u6295\u7a3f\u90ae\u7bb1\uff1aliyazhou@jiqizhixin.com\uff1bzhaoyunfeng@jiqizhixin.com<\/section>\n<\/section>\n<\/blockquote>\n<section><\/section>\n<ul>\n<li>\n<section><strong>\u5f20\u98a2\u7ee7\uff1a<mark data-type=\"institutions\" data-id=\"4349d6b9-40ff-4e23-a237-f9ace91c6c81\">\u6e05\u534e\u5927\u5b66<\/mark>\u672c\u79d1\u751f\uff0c\u5e08\u4ece\u5510\u5f66\u5d69\u8001\u5e08\uff0c\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e3a\u89c6\u9891\u7406\u89e3\u3002<\/strong><\/section>\n<\/li>\n<li>\n<section><strong>\u738b\u9038\u94a6\uff1a<mark data-type=\"institutions\" data-id=\"4349d6b9-40ff-4e23-a237-f9ace91c6c81\">\u6e05\u534e\u5927\u5b66<\/mark>\u6df1\u5733\u56fd\u9645\u7814\u7a76\u751f\u9662\u7855\u58eb\u751f\uff0c\u5e08\u4ece\u5510\u5f66\u5d69\u8001\u5e08\uff0c\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e3a<mark data-type=\"tech_methods\" data-id=\"4468592f-93e9-4575-be91-fd64c0c6afe0\">\u591a\u6a21\u6001\u5b66\u4e60<\/mark>\u3002<\/strong><\/section>\n<\/li>\n<li>\n<section><strong>\u5510\u5f66\u5d69\u535a\u58eb\uff1a<mark data-type=\"institutions\" data-id=\"4349d6b9-40ff-4e23-a237-f9ace91c6c81\">\u6e05\u534e\u5927\u5b66<\/mark>\u6df1\u5733\u56fd\u9645\u7814\u7a76\u751f\u9662\u52a9\u7406\u6559\u6388\uff0c\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e3a<mark data-type=\"tech_tasks\" data-id=\"6e614199-9e49-450e-9078-61fb2b122da9\">\u8ba1\u7b97\u673a\u89c6\u89c9<\/mark>\u4e0e<mark data-type=\"tech_tasks\" data-id=\"4676abd4-83e5-47a9-a955-40fe94535398\">\u6a21\u5f0f\u8bc6\u522b<\/mark>\u3002<\/strong><\/section>\n<\/li>\n<li>\n<section><strong>\u5218\u955b\uff1a<mark data-type=\"institutions\" data-id=\"4349d6b9-40ff-4e23-a237-f9ace91c6c81\">\u6e05\u534e\u5927\u5b66<\/mark>\u6df1\u5733\u56fd\u9645\u7814\u7a76\u751f\u9662\u535a\u58eb\u751f\uff0c\u5e08\u4ece\u5510\u5f66\u5d69\u8001\u5e08\uff0c\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e3a\u89c6\u89c9\u5206\u5272\u3001\u89c6\u9891\u7406\u89e3\u3002<\/strong><\/section>\n<\/li>\n<li>\n<section><strong>\u51af\u4f73\u65f6\u535a\u58eb\uff1a<mark data-type=\"institutions\" data-id=\"7092494f-e4a9-4175-8efe-b5ddb66a9878\">\u5b57\u8282\u8df3\u52a8<\/mark>\u89c6\u89c9\u7814\u7a76\u7684\u8d1f\u8d23\u4eba\uff0c\u4ed6\u7684\u7814\u7a76\u9886\u57df\u5305\u62ec<mark data-type=\"tech_methods\" data-id=\"01946acc-d031-4c0e-909c-f062643b7273\">\u6df1\u5ea6\u5b66\u4e60<\/mark>\u53ca\u5176\u5728<mark data-type=\"tech_tasks\" data-id=\"6e614199-9e49-450e-9078-61fb2b122da9\">\u8ba1\u7b97\u673a\u89c6\u89c9<\/mark>\u4e2d\u7684\u5e94\u7528\u3002<\/strong><\/section>\n<\/li>\n<li>\n<section><strong>\u4ee3\u5b63\u5cf0\u535a\u58eb\uff1a<mark data-type=\"institutions\" data-id=\"4349d6b9-40ff-4e23-a237-f9ace91c6c81\">\u6e05\u534e\u5927\u5b66<\/mark>\u7535\u5b50\u5de5\u7a0b\u7cfb\u526f\u6559\u6388\uff0c\u4e3b\u8981\u7814\u7a76\u65b9\u5411\u4e3a\u89c6\u89c9\u57fa\u7840\u6a21\u578b\u4e0e\u81ea\u52a8\u9a7e\u9a76\u3002<\/strong><\/section>\n<\/li>\n<li>\n<section><strong>\u9879\u76ee Leader\uff0c\u9773\u6f47\u6770\u535a\u58eb\uff1a\u73b0\u5c31\u804c\u4e8e<mark data-type=\"institutions\" data-id=\"7092494f-e4a9-4175-8efe-b5ddb66a9878\">\u5b57\u8282\u8df3\u52a8<\/mark>\u7f8e\u56fd\u7814\u7a76\u9662\uff0c\u7814\u7a76\u65b9\u5411\u5305\u62ec\u591a\u6a21\u6001\u57fa\u7840\u6a21\u578b\u3001\u751f\u6210\u5f0f\u5b66\u4e60\u3001\u89c6\u9891\u7f16\u8f91\u7b49\u3002<\/strong><\/section>\n<\/li>\n<\/ul>\n<section><\/section>\n<section>\u57fa\u4e8e ChatGPT\u3001LLAMA\u3001Vicuna [1, 2, 3] \u7b49\u5927<mark data-type=\"tech_tasks\" data-id=\"fee178e8-ee20-42fc-8f8f-1d41c1f34e8f\">\u8bed\u8a00\u6a21\u578b<\/mark>\uff08Large Language Models\uff0cLLMs\uff09\u7684\u5f3a\u5927\u7406\u89e3\u3001\u751f\u6210\u548c\u63a8\u7406\u80fd\u529b\uff0c\u591a\u6a21\u6001\u5927\u6a21\u578b\uff08Large Multimodal Models\uff0cLMMs\uff09\u5728\u56fe\u7247\u89c6\u89c9\u7406\u89e3\u4efb\u52a1\u4e0a\u53d6\u5f97\u4e86\u6210\u529f\uff0c\u5982 MiniGPT-4\u3001LLAVA [4, 5, 6] \u7b49\u7b49\u3002\u66f4\u8fdb\u4e00\u6b65\u5730\uff0c\u4e00\u4e9b\u5de5\u4f5c\u5c06 LMM \u5f3a\u5927\u7684\u56fe\u7247\u7406\u89e3\u80fd\u529b\u8fc1\u79fb\u5230\u89c6\u9891\u9886\u57df\uff0c\u4f7f\u5f97\u89c6\u9891\u5185\u5bb9\u7406\u89e3\u548c\u63a8\u7406\u6210\u4e3a\u53ef\u80fd\uff0c\u4f8b\u5982 Video-ChatGPT\u3001Vista-LLaMA [7, 8] \u7b49\u3002<\/section>\n<section><\/section>\n<section>\u7136\u800c\uff0c\u5927\u591a\u6570\u591a\u6a21\u6001\u6a21\u578b<strong>\u4ec5\u80fd\u5bf9\u8f83\u77ed\u7684\u79bb\u7ebf\u89c6\u9891\u6570\u636e\u8fdb\u884c\u6587\u672c\u63cf\u8ff0\u6216\u95ee\u7b54\uff0c\u5bf9\u4e8e\u957f\u89c6\u9891\u548c\u5728\u7ebf\u89c6\u9891\u6d41\u7684\u7406\u89e3\u80fd\u529b\u6bd4\u8f83\u6709\u9650<\/strong>\u3002\u8ba9\u6a21\u578b\u5177\u6709\u7406\u89e3\u957f\u89c6\u9891\u7684\u80fd\u529b\u662f\u901a\u5f80\u66f4\u667a\u80fd\u7684\u6a21\u578b\u751a\u81f3\u8fbe\u5230 AGI \u7684\u8def\u5f84\u3002\u8fd9\u4e00\u7814\u7a76\u7a7a\u767d\u9650\u5236\u4e86\u591a\u6a21\u6001\u5927\u6a21\u578b\u5728\u8bb8\u591a\u5728\u7ebf\u573a\u666f\u4e2d\u7684\u5b9e\u9645\u5e94\u7528\uff0c\u5982\u5177\u8eab<mark data-type=\"concepts\" data-id=\"2d28aa9c-942d-471d-bd96-8bfefb7144e0\">\u4eba\u5de5\u667a\u80fd<\/mark>\u3001\u667a\u80fd\u76d1\u63a7\u7cfb\u7edf\u7b49\u3002<\/section>\n<section><\/section>\n<section>\u9488\u5bf9\u8fd9\u70b9\uff0c\u4e00\u4e9b\u5de5\u4f5c [9, 10] \u5f00\u59cb\u7814\u7a76\u5982\u4f55\u589e\u5f3a\u5bf9\u957f\u89c6\u9891\u7684\u7406\u89e3\u80fd\u529b\uff0c\u5927\u591a\u57fa\u4e8e\u5e27\u91c7\u6837\u548c\u7279\u5f81\u878d\u5408\u7684\u65b9\u6cd5\u3002\u7136\u800c\uff0c\u73b0\u6709\u7684\u65b9\u6cd5\u5b58\u5728\u4ee5\u4e0b\u7f3a\u70b9\uff1a1)<strong>\u00a0\u663e\u5b58\u5f00\u9500\u548c\u56de\u7b54\u5ef6\u8fdf\u968f\u8f93\u5165\u5e27\u6570\u91cf\u589e\u957f<\/strong>\uff0c\u8fd9\u4e3a\u957f\u89c6\u9891\u7406\u89e3\u5e26\u6765\u56f0\u96be\uff0c\u53ea\u80fd\u4f7f\u7528\u7a00\u758f\u91c7\u6837\u7b49\u65b9\u5f0f\uff0c\u800c\u8fd9\u4f1a\u663e\u8457\u5f71\u54cd\u6a21\u578b\u6027\u80fd\u30022)\u00a0<strong>\u65e0\u6cd5\u5904\u7406\u5728\u7ebf\u89c6\u9891\u6d41<\/strong>\uff0c\u53ea\u80fd\u5c06\u5728\u7ebf\u89c6\u9891\u6d41\u8fdb\u884c\u5206\u6bb5\u5904\u7406\uff0c\u96be\u4ee5\u5904\u7406\u65b0\u8f93\u5165\u7684\u89c6\u9891\u7247\u6bb5\u4e0e\u65e7\u89c6\u9891\u7247\u6bb5\u4e4b\u95f4\u7684\u4fe1\u606f\u4ea4\u4e92\uff0c\u963b\u788d\u4e86 LMM \u5bf9\u957f\u89c6\u9891\u6d41\u6574\u4f53\u7684\u7406\u89e3\u80fd\u529b\u3002<\/section>\n<section><\/section>\n<section>\u4e3a\u4e86\u89e3\u51b3\u6b64\u95ee\u9898\uff0c<mark data-type=\"institutions\" data-id=\"7092494f-e4a9-4175-8efe-b5ddb66a9878\">\u5b57\u8282\u8df3\u52a8<\/mark>\u8054\u5408<mark data-type=\"institutions\" data-id=\"4349d6b9-40ff-4e23-a237-f9ace91c6c81\">\u6e05\u534e\u5927\u5b66<\/mark>\u7684\u7814\u7a76\u4eba\u5458\u4eff\u7167\u4eba\u7c7b\u7684<mark data-type=\"concepts\" data-id=\"1ccbfcc7-1b56-44d6-b64b-729d855abcb1\">\u611f\u77e5<\/mark>\u548c\u8bb0\u5fc6\u673a\u5236\uff0c\u63d0\u51fa\u4e86<strong>\u9996\u4e2a\u9488\u5bf9\u957f\u89c6\u9891\u6d41\u7684\u5728\u7ebf\u7406\u89e3\u591a\u6a21\u6001\u5927\u6a21\u578b Flash-VStream<\/strong>\u3002<\/section>\n<section><\/section>\n<section>\u5728\u5177\u4f53\u4ecb\u7ecd\u5b83\u4e4b\u524d\uff0c\u5148\u6765\u4f53\u9a8c\u4e00\u4e0b Flash-VStream \u7684\u5b9e\u65f6\u95ee\u7b54\u80fd\u529b\uff1a<a href=\"https:\/\/mp.weixin.qq.com\/s\/O0Ao-58yTblQb6ODaWyssg\"><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-883ffeb654d8f39f3ab61930eac99e19.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-883ffeb654d8f39f3ab61930eac99e19.png\"><\/a><\/a><\/section>\n<section>\u6211\u4eec\u53ef\u4ee5\u770b\u5230\u6a21\u578b\u5bf9\u957f\u89c6\u9891\u4e0a\u4e0b\u6587\u6709\u6bd4\u8f83\u597d\u7684\u8bb0\u5fc6\u80fd\u529b\uff0c\u80fd\u591f\u7ed9\u51fa\u7b26\u5408\u89c6\u9891\u60c5\u666f\u7684\u56de\u590d\u3002\u4f8b\u5982\u5728 56:00 \u65f6\u523b\u63d0\u95ee\u6293\u53d6\u9762\u7c89\uff08\u53d1\u751f\u5728\u5341\u51e0\u5206\u949f\u4e4b\u524d\uff09\u4e4b\u540e\u4e3b\u4eba\u516c\u505a\u4e86\u4ec0\u4e48\u52a8\u4f5c\uff0c\u6a21\u578b\u80fd\u591f\u8fc5\u901f\u7ed9\u51fa\u6b63\u786e\u800c\u8be6\u7ec6\u7684\u56de\u7b54\u3002<strong>Flash-VStream \u6a21\u578b\u80fd\u591f\u5904\u7406\u9488\u5bf9\u5927\u65f6\u95f4\u8de8\u5ea6\u7684\u89c6\u9891\u95ee\u9898\uff0c\u53cd\u6620\u4e86\u6a21\u578b\u5177\u6709\u9ad8\u6548\u8bb0\u5fc6\u957f\u89c6\u9891\u89c6\u89c9\u4fe1\u606f\u7684\u80fd\u529b<\/strong>\u3002<\/section>\n<section><\/section>\n<section>\u76f8\u6bd4\u4e4b\u524d\u7684\u5de5\u4f5c\uff0cFlash-VStream \u7684\u4f18\u52bf\u5728\u4e8e\uff1a<\/section>\n<section><\/section>\n<ul>\n<li>\n<section>\u80fd\u591f<strong>\u5728\u7ebf\u5904\u7406\u6781\u957f\u7684\u89c6\u9891<mark data-type=\"tech_methods\" data-id=\"8a259fbf-2436-4224-8156-2a67068cadc3\">\u6d41\u6570\u636e<\/mark>\uff0c\u5feb\u901f\u8bb0\u5fc6\u91cd\u8981\u4fe1\u606f\uff0c\u5b9e\u65f6\u56de\u7b54\u7528\u6237\u63d0\u95ee<\/strong>\u3002<\/section>\n<\/li>\n<li>\n<section>\u968f\u7740\u8f93\u5165\u5e27\u6570\u91cf\u7684\u589e\u52a0\uff0c<strong>\u663e\u5b58\u5f00\u9500\u548c\u56de\u7b54\u5ef6\u8fdf\u51e0\u4e4e\u6ca1\u6709\u53d8\u5316<\/strong>\uff0c\u5b9e\u73b0\u4e86\u9ad8\u6548\u7684\u957f\u89c6\u9891\u7406\u89e3\u3002<\/section>\n<\/li>\n<li>\n<section>\u5229\u7528 STAR \u8bb0\u5fc6\u673a\u5236<strong>\u5bf9\u4e0d\u540c\u7c92\u5ea6\u8bed\u4e49\u4fe1\u606f\u8fdb\u884c\u9ad8\u6548\u878d\u5408<\/strong>\uff0c\u5728\u591a\u4e2a\u957f\u89c6\u9891\u95ee\u7b54 benchmark \u4e0a\u8fbe\u5230 SOTA\u3002<\/section>\n<\/li>\n<\/ul>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-d8ea5dc487b9858a069570569d156dd8.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-d8ea5dc487b9858a069570569d156dd8.png\"><\/a><\/p>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-85cb944e44fc93c18a49d8cb840c8af1.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-85cb944e44fc93c18a49d8cb840c8af1.png\"><\/a><\/p>\n<section>Flash-VStream \u4e0d\u4ec5\u5728\u591a\u4e2a\u957f\u89c6\u9891\u7406\u89e3 benchmark \u4e0a\u8868\u73b0\u4f18\u79c0\uff0c\u8fd8\u83b7\u5f97\u4e86 <strong>CVPR&#8217;24 \u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b<\/strong> Long-Term Video Question Answering Challenge @ CVPR 2024 Workshop \u7684<strong>\u51a0\u519b<\/strong>\u3002<\/section>\n<section><\/section>\n<section>\u5730\u5740\uff1ahttps:\/\/sites.google.com\/view\/loveucvpr24\/track1<\/section>\n<section><\/section>\n<section>\u66f4\u8fdb\u4e00\u6b65\uff0c\u4e3a\u4e86\u652f\u6301\u8fd9\u4e00\u7814\u7a76\u9886\u57df\u7684\u6a21\u578b\u8bc4\u4ef7\u548c\u6539\u8fdb\uff0c\u7814\u7a76\u56e2\u961f\u5728 Ego4D [11] \u548c Movienet [12] \u7684\u57fa\u7840\u4e0a\uff0c\u501f\u52a9 GPT-4V \u6784\u5efa\u4e86\u4e00\u4e2a\u9762\u5411\u5728\u7ebf\u89c6\u9891\u6d41\u95ee\u7b54\u573a\u666f\u7684\u6570\u636e\u96c6 VStream-QA\uff0c\u5b83\u5305\u542b\u603b\u8ba1 21h \u7684\u89c6\u9891\uff0c\u5e73\u5747\u957f\u5ea6\u4e3a 40min\uff0c\u6bcf\u4e2a\u95ee\u7b54\u5bf9\u90fd\u57fa\u4e8e\u7279\u5b9a\u7684\u5df2\u6807\u6ce8\u7684\u65f6\u95f4\u533a\u95f4\u3002\u5728\u8bc4\u4ef7\u65f6\uff0c\u8981\u6c42\u6a21\u578b\u5728\u591a\u4e2a\u65f6\u95f4\u70b9\uff0c\u57fa\u4e8e\u5230\u5f53\u65f6\u523b\u4e3a\u6b62\u7684\u89c6\u9891\u7247\u6bb5\u56de\u7b54\u95ee\u9898\u3002<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-a150d89fbea8078cf7b9290457f1fa69.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-a150d89fbea8078cf7b9290457f1fa69.png\"><\/a><\/p>\n<ul>\n<li>\n<section>\u9879\u76ee\u4e3b\u9875\uff1ahttps:\/\/invinciblewyq.github.io\/vstream-page<\/section>\n<\/li>\n<li>\n<section>\u8bba\u6587\u94fe\u63a5\uff1ahttps:\/\/arxiv.org\/abs\/2406.08085<\/section>\n<\/li>\n<li>\n<section>\u4ee3\u7801\u4ed3\u5e93\uff1ahttps:\/\/github.com\/IVGSZ\/Flash-VStream<\/section>\n<\/li>\n<li>\n<section>\u5728\u7ebf\u4f53\u9a8c\uff1ahttps:\/\/huggingface.co\/spaces\/IVGSZ\/Flash-VStream-demo<\/section>\n<\/li>\n<\/ul>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-b7ed4522def2fc8639c29a15d602b174.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-b7ed4522def2fc8639c29a15d602b174.png\"><\/a><\/p>\n<section>\u4e0d\u540c\u4e8e\u4f20\u7edf\u89c6\u9891\u7406\u89e3 LMM\uff0cFlash-VStream \u5c06\u89c6\u89c9\u4fe1\u606f<mark data-type=\"concepts\" data-id=\"1ccbfcc7-1b56-44d6-b64b-729d855abcb1\">\u611f\u77e5<\/mark>\u8bb0\u5fc6\u548c\u95ee\u7b54\u4ea4\u4e92\u89e3\u8026\uff0c\u4f7f\u7528\u591a\u8fdb\u7a0b\u7cfb\u7edf\u5b9e\u73b0\u4e86\u5bf9\u957f\u89c6\u9891\u6d41\u7684\u5b9e\u65f6\u5904\u7406\u3002\u90a3\u4e48\u8fd9\u9879\u7814\u7a76\u5177\u4f53\u662f\u5982\u4f55\u505a\u7684\u5462\uff1f<\/section>\n<section><\/section>\n<section><strong>\u6a21\u578b\u6838\u5fc3\uff1aSTAR \u8bb0\u5fc6\u673a\u5236<\/strong><\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-653417d54f50f59f9e3bd0f692729c9e.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-653417d54f50f59f9e3bd0f692729c9e.png\"><\/a><\/p>\n<section>\u5982\u8bba\u6587\u4e2d\u7684\u6846\u67b6\u56fe\u6240\u793a\uff0cFlash-VStream \u67b6\u6784\u5341\u5206\u7b80\u6d01\uff0c\u7531\u5e27\u5904\u7406\u8fdb\u7a0b\u548c\u95ee\u9898\u5904\u7406\u8fdb\u7a0b\u7ec4\u6210\uff0c\u5176\u6a21\u578b\u5305\u62ec\u56db\u4e2a\u4e3b\u8981\u90e8\u5206\uff1a1) \u9884\u8bad\u7ec3\u7684 CLIP-ViT \u89c6\u89c9\u7f16\u7801\u5668\uff1b2) \u5927<mark data-type=\"tech_tasks\" data-id=\"fee178e8-ee20-42fc-8f8f-1d41c1f34e8f\">\u8bed\u8a00\u6a21\u578b<\/mark>\uff1b3\uff09STAR \u8bb0\u5fc6\u673a\u5236\uff1b4\uff09\u7279\u5f81\u7f13\u51b2\u533a\u3002\u5176\u4e2d\uff0c\u540e\u4e24\u8005\u662f Flash-VStream \u7684\u6838\u5fc3\u3002STAR \u8bb0\u5fc6\u5305\u62ec \u201c\u7a7a\u95f4\u201d\u3001\u201c\u65f6\u95f4\u201d\u3001\u201c\u62bd\u8c61\u201d\u3001\u201c\u68c0\u7d22\u201d \u56db\u79cd\u8bb0\u5fc6\u6a21\u5757\uff0c\u7528\u4e8e\u9ad8\u6548\u878d\u5408\u4e0d\u540c\u7c92\u5ea6\u7684\u8bed\u4e49\u4fe1\u606f\uff0c\u5b9e\u73b0\u4e86\u5e27\u7ea7\u522b\u7684\u4fe1\u606f\u805a\u5408\u3002\u7279\u5f81\u7f13\u51b2\u533a\u8f85\u52a9\u68c0\u7d22\u8bb0\u5fc6\uff0c\u7c7b\u4f3c\u4e8e\u4eba\u7c7b\u56de\u5fc6\u8d77\u5370\u8c61\u6df1\u523b\u7684\u4e8b\u4ef6\u4e00\u6837\uff0c\u4ece\u5386\u53f2\u89c6\u9891\u4e2d\u68c0\u7d22\u51fa\u5173\u952e\u4fe1\u606f\uff0c\u4ee5\u63d0\u9ad8\u6a21\u578b\u5bf9\u957f\u89c6\u9891\u4e2d\u91cd\u8981\u4e8b\u4ef6\u7ec6\u8282\u7684\u7406\u89e3\u80fd\u529b\u3002<\/section>\n<section><\/section>\n<section>\u5176\u4e2d\uff0c\u7a7a\u95f4\u8bb0\u5fc6\u548c\u68c0\u7d22\u8bb0\u5fc6\u6bcf\u5e27\u5177\u6709\u6700\u591a\u7684 token \u6570\u91cf\uff0c\u65f6\u95f4\u8bb0\u5fc6\u6b21\u4e4b\uff0c\u62bd\u8c61\u8bb0\u5fc6\u6bcf\u5e27\u4ec5\u7528 1 \u4e2a token \u8868\u793a\u3002\u8fd9\u79cd\u8bbe\u8ba1\u9ad8\u6548\u8868\u793a\u4e86\u4ece\u6700\u5177\u4f53\u5230\u6700\u62bd\u8c61\u7684\u89c6\u89c9\u7279\u5f81\u3002\u4e3a\u4e86\u5f97\u5230\u66f4\u5c0f\u7684\u7279\u5f81\u56fe\uff0cFlash-VStream \u5728\u7a7a\u95f4\u7ef4\u5ea6\u4f7f\u7528\u5e73\u5747<mark data-type=\"concepts\" data-id=\"0a4cedf0-0ee0-4406-946e-2877950da91d\">\u6c60\u5316<\/mark>\u64cd\u4f5c\u3002<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-9397cfbe0239e9e4baa6d3a8cc9d88c8.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-9397cfbe0239e9e4baa6d3a8cc9d88c8.png\"><\/a><\/p>\n<section>\u6839\u636e\u7814\u7a76\u4eba\u5458\u7684\u63cf\u8ff0\uff0cSTAR \u8bb0\u5fc6\u91c7\u7528\u4e86\u56db\u79cd\u7b80\u6d01\u9ad8\u6548\u7684\u8bb0\u5fc6\u66f4\u65b0\u673a\u5236\uff1a<\/section>\n<section><\/section>\n<ul>\n<li>\n<section>\u5bf9\u4e8e\u7a7a\u95f4\u8bb0\u5fc6\u548c\u7279\u5f81\u7f13\u51b2\u533a\uff0c\u901a\u8fc7 FIFO\uff08First-In-First-Out\uff09\u961f\u5217\u66f4\u65b0\u3002\u961f\u5217\u7ef4\u62a4\u4e86\u6700\u65b0\u7684\u82e5\u5e72\u5e27\uff0c\u786e\u4fdd\u6a21\u578b\u5bf9\u6700\u65b0\u7684\u7ec6\u7c92\u5ea6\u7a7a\u95f4\u4fe1\u606f\u6709\u8f83\u5f3a\u7684<mark data-type=\"concepts\" data-id=\"1ccbfcc7-1b56-44d6-b64b-729d855abcb1\">\u611f\u77e5<\/mark>\u80fd\u529b\u3002<\/section>\n<\/li>\n<li>\n<section>\u5bf9\u4e8e\u65f6\u95f4\u8bb0\u5fc6\uff0c\u5f53\u8f93\u5165 token \u6570\u91cf\u8d85\u8fc7\u8bb0\u5fc6\u5bb9\u91cf\u65f6\uff0c\u91c7\u7528\u52a0\u6743 K-means <mark data-type=\"concepts\" data-id=\"de069734-8912-43fb-b365-3b014165bc4d\">\u805a\u7c7b<\/mark>\u7b97\u6cd5\u8fdb\u884c\u5e27\u7ea7\u522b\u7684\u7279\u5f81\u805a\u5408\u3002\u8be5\u7b97\u6cd5\u5c06\u65f6\u95f4\u8bb0\u5fc6\u7684 token \u548c\u65b0\u8f93\u5165\u7684 token \u4e00\u8d77<mark data-type=\"concepts\" data-id=\"de069734-8912-43fb-b365-3b014165bc4d\">\u805a\u7c7b<\/mark>\u4e3a\u4e00\u4e9b\u7c07\uff08\u7c07\u7684\u6570\u91cf\u5c31\u662f\u8bb0\u5fc6\u5bb9\u91cf\uff0c\u7c07\u7684\u5927\u5c0f\u662f\u5176\u6240\u5305\u542b\u5e27\u7684\u6570\u91cf\uff09\uff0c\u5e76\u7528\u8fd9\u4e9b\u7c07\u7684\u8d28\u5fc3\u4f5c\u4e3a\u65b0\u8bb0\u5fc6\uff0c\u4ee3\u8868\u76f8\u5e94\u7684\u5173\u952e\u4e8b\u4ef6\u4fe1\u606f\u3002\u8fd9\u79cd\u65b9\u6cd5\u53ef\u4ee5\u7b80\u6d01\u9ad8\u6548\u5730\u5b58\u50a8\u65f6\u5e8f\u76f8\u5173\u7684\u4e0a\u4e0b\u6587\u4fe1\u606f\u3002<\/section>\n<\/li>\n<li>\n<section>\u5bf9\u4e8e\u62bd\u8c61\u8bb0\u5fc6\uff0c\u5f15\u5165\u4e86\u8bed\u4e49\u6ce8\u610f\u529b\u6a21\u578b\uff08Semantic Attention\uff09\uff0c\u5c06\u7a7a\u95f4\u4e0e\u65f6\u95f4\u7279\u5f81\u62bd\u8c61\u6210\u6700\u9ad8\u5c42\u6b21\u7684\u8bed\u4e49\u7279\u5f81\u3002\u8be5\u6a21\u578b\u7528\u57fa\u4e8e\u6ce8\u610f\u529b\u548c<mark data-type=\"concepts\" data-id=\"d6ee5e5b-43ff-4c41-87ff-f34c234d0e32\">\u52a8\u91cf<\/mark>\u7684\u65b9\u5f0f\u66f4\u65b0\u62bd\u8c61\u8bb0\u5fc6\uff0c\u4f7f\u5176\u59cb\u7ec8\u8868\u793a\u89c6\u9891\u7ea7\u522b\u7684\u9ad8\u5c42\u6b21\u8bed\u4e49\u4fe1\u606f\u3002<\/section>\n<\/li>\n<li>\n<section>\u5bf9\u4e8e\u68c0\u7d22\u8bb0\u5fc6\uff0c\u901a\u8fc7\u8bc6\u522b\u5173\u952e\u5e27\u7279\u5f81\u8fdb\u884c\u66f4\u65b0\u3002\u9996\u5148\u4ece\u65f6\u95f4\u8bb0\u5fc6\u4e2d\u9009\u62e9\u51fa\u6700\u5927\u7684\u82e5\u5e72\u7c07\uff0c\u7136\u540e\u4ece\u7279\u5f81\u7f13\u51b2\u533a\u4e2d\u68c0\u7d22\u51fa\u4e0e\u8fd9\u4e9b\u7c07\u7684\u8d28\u5fc3 L2 \u8ddd\u79bb\u6700\u8fd1\u7684\u5e27\u7684\u7279\u5f81\uff0c\u4ee5\u6b64\u4f5c\u4e3a\u5173\u952e\u4e8b\u4ef6\u7684\u56de\u5fc6\uff0c\u4e3a\u65f6\u95f4\u8bb0\u5fc6\u8865\u5145\u76f8\u5e94\u7684\u7ec6\u7c92\u5ea6\u4fe1\u606f\u3002<\/section>\n<\/li>\n<\/ul>\n<section><\/section>\n<section>Flash-VStream \u51ed\u501f\u5176\u521b\u65b0\u6027\u7684 STAR \u8bb0\u5fc6\u673a\u5236\uff0c\u4e0d\u4ec5\u80fd\u591f\u9ad8\u6548\u878d\u5408\u4e0d\u540c\u7c92\u5ea6\u7684\u8bed\u4e49\u4fe1\u606f\uff0c\u8fd8\u80fd\u901a\u8fc7\u7279\u5f81\u7f13\u51b2\u533a\u7684\u8f85\u52a9\uff0c\u7cbe\u786e\u5730\u56de\u5fc6\u548c\u68c0\u7d22\u957f\u89c6\u9891\u4e2d\u91cd\u8981\u4e8b\u4ef6\u7684\u7ec6\u8282\u4fe1\u606f\uff0c\u4ece\u800c\u663e\u8457\u63d0\u5347\u6a21\u578b\u7684\u7406\u89e3\u80fd\u529b\u4e0e\u6027\u80fd\u3002<\/section>\n<section><\/section>\n<section><strong>VStream-QA \u6570\u636e\u96c6<\/strong><\/section>\n<section><\/section>\n<section>\u6709\u4e86\u4e0a\u8ff0\u5b9e\u73b0\u65b9\u6848\uff0c\u8fd8\u9700\u8981\u6709\u5408\u9002\u7684\u6d4b\u8bd5\u6570\u636e\u6765\u8bc4\u4ef7\u6a21\u578b\u5bf9\u5728\u7ebf\u89c6\u9891\u6d41\u7684\u7406\u89e3\u80fd\u529b\u3002\u56de\u987e\u73b0\u6709\u7684\u957f\u89c6\u9891\u95ee\u7b54\u6570\u636e\u96c6\uff0c\u5b83\u4eec\u7684\u4e3b\u8981\u76ee\u7684\u5927\u591a\u662f\u8bc4\u4ef7\u6a21\u578b\u7684\u63cf\u8ff0\u6027\u95ee\u7b54\u80fd\u529b\u3001\u65f6\u5e8f\u7406\u89e3\u80fd\u529b\u3001\u7535\u5f71\u7406\u89e3\u80fd\u529b\u7b49\uff0c\u5747\u5c5e\u4e8e\u79bb\u7ebf\u7406\u89e3\u80fd\u529b\u3002\u5e76\u4e14\u5b83\u4eec\u7684\u89c6\u9891\u5e73\u5747\u957f\u5ea6\u5c40\u9650\u5728 4 \u5206\u949f\u4ee5\u5185\u3002<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-511d8599dfbb705ca70d4533ba190688.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-511d8599dfbb705ca70d4533ba190688.png\"><\/a><\/p>\n<section>\u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e9b\u95ee\u9898\uff0c\u7814\u7a76\u56e2\u961f\u7b5b\u9009\u4e86 Ego4d \u548c Movienet \u4e2d\u7684\u4e00\u90e8\u5206\u89c6\u9891\u7247\u6bb5\uff0c\u4e3a\u6bcf\u4e2a\u89c6\u9891\u7247\u6bb5\u6807\u6ce8\u4e86\u591a\u4e2a\u95ee\u7b54\u5bf9\uff0c\u5e76\u6807\u8bb0\u4e86\u7b54\u6848\u6240\u5728\u7684\u89c6\u9891\u533a\u95f4\u3002\u5728\u6d4b\u8bd5\u65f6\uff0c\u8981\u6c42\u6a21\u578b\u5728\u591a\u4e2a\u65f6\u95f4\u70b9\uff0c\u57fa\u4e8e\u5230\u5f53\u65f6\u523b\u4e3a\u6b62\u7684\u89c6\u9891\u7247\u6bb5\u56de\u7b54\u95ee\u9898\uff0c\u4ee5\u6b64\u6d4b\u8bd5\u6a21\u578b\u7684\u5728\u7ebf\u89c6\u9891\u6d41\u7406\u89e3\u80fd\u529b\u3002\u8fd9\u5c31\u662f VStream-QA \u6570\u636e\u96c6\uff0c\u5176\u6837\u4f8b\u5982\u4e0b\u56fe\u6240\u793a\uff1a<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-28dda841f616d2b84a930cb4455db22a.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-28dda841f616d2b84a930cb4455db22a.png\"><\/a><\/p>\n<section>\u548c\u4e3b\u6d41\u7684\u5f00\u653e\u8bcd\u5178\u79bb\u7ebf\u89c6\u9891\u95ee\u7b54\u6570\u636e\u96c6\u76f8\u540c\uff0cVStream-QA \u6570\u636e\u96c6\u4e5f\u91c7\u7528\u57fa\u4e8e GPT-3.5 \u7684\u8bc4\u4ef7\u6307\u6807\u3002\u5177\u4f53\u6765\u8bf4\uff0c\u5411 GPT-3.5 \u8f93\u5165\u95ee\u9898\u3001\u6807\u51c6\u7b54\u6848\u3001\u6a21\u578b\u7684\u9884\u6d4b\u4e09\u5143\u7ec4\uff0c\u7531 GPT \u6a21\u578b\u6765\u5224\u65ad\u8be5\u7b54\u6848\u662f\u5426\u51c6\u786e\u56de\u7b54\uff0c\u4ee5\u53ca\u53ef\u4fe1\u5ea6\u5206\u6570\u662f\u591a\u5c11\u3002\u7edf\u8ba1\u6240\u6709\u95ee\u9898\u7684\u6307\u6807\u5373\u4e3a<mark data-type=\"concepts\" data-id=\"8be77eae-12da-4e9e-9a88-b7f5bae98c2e\">\u51c6\u786e\u7387<\/mark>\uff08Acc.\uff09\u548c\u53ef\u4fe1\u5ea6\u5206\u6570\uff08Sco.\uff09\u3002<\/section>\n<section><\/section>\n<section><strong>\u7b97\u6cd5\u6d4b\u8bc4<\/strong><\/section>\n<section><\/section>\n<section>\u7814\u7a76\u56e2\u961f\u5728\u65b0\u63d0\u51fa\u7684\u5728\u7ebf\u89c6\u9891\u6d41\u95ee\u7b54 Real-time VStream-QA Benchmark \u4e0a\u8bc4\u6d4b\u4e86 Flash-VStream \u7684\u5b9e\u65f6\u89c6\u9891\u7406\u89e3\u6027\u80fd\uff0c\u5305\u62ec RVS-Ego \u548c RVS-Movie \u4e24\u4e2a\u5b50\u96c6\u3002\u5f97\u76ca\u4e8e STAR \u8bb0\u5fc6\u673a\u5236\u7684\u9ad8\u6548\u8bbe\u8ba1\uff0cFlash-VStream \u5177\u6709\u6781\u4f4e\u7684\u56de\u7b54\u5ef6\u8fdf\u548c\u663e\u5b58\u5360\u7528\uff0c\u5e76\u4e14\u51e0\u4e4e\u4e0d\u968f\u8f93\u5165\u5e27\u7684\u6570\u91cf\u53d8\u5316\uff0c\u4e3a\u5b9e\u65f6\u95ee\u7b54\u7684\u6027\u80fd\u63d0\u4f9b\u4fdd\u969c\u3002<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-033c2085bf11f284930939f3ffe17c31.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-033c2085bf11f284930939f3ffe17c31.png\"><\/a><\/p>\n<section>\u540c\u65f6\uff0c\u4e3a\u4e86\u8bc4\u4ef7 Flash-VStream \u6a21\u578b\u5bf9\u4e8e\u79bb\u7ebf\u89c6\u9891\u7684\u7406\u89e3\u80fd\u529b\uff0c\u7814\u7a76\u56e2\u961f\u5728\u56db\u4e2a\u79bb\u7ebf\u89c6\u9891\u95ee\u7b54 Benchmark \u4e0a\u8bc4\u6d4b\u4e86 Flash-VStream \u7684\u89c6\u9891\u7406\u89e3\u6027\u80fd\u3002\u6b64\u5916\uff0c\u8fd8\u5728\u79bb\u7ebf\u7248 VStream-QA \u6570\u636e\u96c6\u8fdb\u884c\u4e86\u6d4b\u8bd5\uff0c\u5206\u4e3a VS-Ego \u548c VS-Movie \u4e24\u4e2a\u5b50\u96c6\u3002\u79bb\u7ebf\u7248 VStream-QA \u6570\u636e\u96c6\u9488\u5bf9\u6bcf\u4e2a\u95ee\u9898\uff0c\u53ea\u8f93\u5165\u8be5\u95ee\u9898\u7b54\u6848\u6240\u5728\u7684\u89c6\u9891\u7247\u6bb5\u5e76\u8fdb\u884c\u63d0\u95ee\uff0c\u76f8\u6bd4\u4e8e\u5728\u7ebf\u7248 Real-time VStream-QA \u96be\u5ea6\u8f83\u4f4e\u3002<\/section>\n<p><a href=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-70eaf7ec54fd84949b0e5d454dbc1ac9.png\" data-fancybox=\"images\" data-fancybox=\"gallery\"><img decoding=\"async\" src=\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-70eaf7ec54fd84949b0e5d454dbc1ac9.png\"><\/a><\/p>\n<section>\u5728\u516d\u4e2a benchmark \u7684<mark data-type=\"concepts\" data-id=\"8be77eae-12da-4e9e-9a88-b7f5bae98c2e\">\u51c6\u786e\u7387<\/mark>\u548c\u53ef\u4fe1\u5ea6\u5206\u6570\u4e0a\uff0cFlash-VStream \u7684\u6027\u80fd\u5747\u4f18\u4e8e\u5176\u4ed6\u65b9\u6cd5\uff0c\u8bc1\u660e\u5176\u5f3a\u5927\u7684\u79bb\u7ebf\u89c6\u9891\u7406\u89e3\u80fd\u529b\u3002<\/section>\n<section><\/section>\n<section>\u611f\u5174\u8da3\u7684\u5c0f\u4f19\u4f34\u53ef\u4ee5\u5173\u6ce8\u4e00\u6ce2\uff0c\u4ee3\u7801\u5df2\u7ecf\u5f00\u6e90\u5566\uff5e<\/section>\n<section><\/section>\n<section><em><sup>\u76f8\u5173\u6587\u732e\uff1a<\/sup><\/em><\/section>\n<section><sup><em>[1]: OpenAI. Chatgpt: A language model for conversational AI.<\/em><\/sup><\/section>\n<section><sup><em>[2]: Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timoth \u0301ee Lacroix, Baptiste Rozi`ere, Naman Goyal, Eric Hambro, Faisal Azhar, et al. Llama: Open and efficient foundation language models.<\/em><\/sup><\/section>\n<section><sup><em>[3]: Wei-Lin Chiang, Zhuohan Li, Zi Lin, Ying Sheng, Zhanghao Wu, Hao Zhang, Lianmin Zheng, Siyuan Zhuang, Yonghao Zhuang, Joseph E Gonzalez, et al. Vicuna: An open-source chatbot impressing gpt-4 with 90% chatgpt qualit<\/em><\/sup><\/section>\n<section><sup><em>[4]: Deyao Zhu, Jun Chen, Xiaoqian Shen, Xiang Li, and Mohamed Elhoseiny. Minigpt-4: Enhancing vision-language understanding with advanced large language models.<\/em><\/sup><\/section>\n<section><sup><em>[5]: Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning.<\/em><\/sup><\/section>\n<section><sup><em>[6]: Liu, Haotian, Chunyuan Li, Yuheng Li, and Yong Jae Lee. Improved baselines with visual instruction tuning.<\/em><\/sup><\/section>\n<section><sup><em>[7]: Maaz, Muhammad, Hanoona Rasheed, Salman Khan, and Fahad Shahbaz Khan. Video-chatgpt: Towards detailed video understanding via large vision and language models.\u00a0<\/em><\/sup><\/section>\n<section><sup><em>[8]: Ma, Fan, Xiaojie Jin, Heng Wang, Yuchen Xian, Jiashi Feng, and Yi Yang. VISTA-LLAMA: Reducing Hallucination in Video Language Models via Equal Distance to Visual Tokens.<\/em><\/sup><\/section>\n<section><sup><em>[9]: Song, Enxin, Wenhao Chai, Guanhong Wang, Yucheng Zhang, Haoyang Zhou, Feiyang Wu, Haozhe Chi et al. Moviechat: From dense token to sparse memory for long video understanding.<\/em><\/sup><\/section>\n<section><sup><em>[10]: Jin, Peng, Ryuichi Takanobu, Wancai Zhang, Xiaochun Cao, and Li Yuan. Chat-univi: Unified visual representation empowers large language models with image and video understanding.<\/em><\/sup><\/section>\n<section><sup><em>[11]: Grauman, Kristen, Andrew Westbury, Eugene Byrne, Zachary Chavis, Antonino Furnari, Rohit Girdhar, Jackson Hamburger et al. Ego4d: Around the world in 3,000 hours of egocentric video.<\/em><\/sup><\/section>\n<section><em><sup>[12]: Huang, Qingqiu, Yu Xiong, Anyi Rao, Jiaze Wang, and Dahua Lin. Movienet: A holistic dataset for movie understanding.<\/sup><\/em><\/section>\n<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:<a href=\"https:\/\/www.jiqizhixin.com\/articles\/2024-07-08-12\" target=\"_blank\">\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR&#8217;24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u6587\u7ae0\u6765\u6e90\u4e8e\u4e92\u8054\u7f51:\u5f00\u6e90\u89c6\u9891\u7248GPT-4o [&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":[73,68,65,66],"class_list":["post-34453","post","type-post","status-publish","format-standard","hentry","category-news","tag-73","tag-68","tag-65","tag-66"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v26.4 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR&#039;24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b - \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=34453\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/17aitech.com\/?p=34453\",\"url\":\"https:\/\/17aitech.com\/?p=34453\",\"name\":\"\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR'24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b - \u4e00\u8d77AI\u6280\u672f\",\"isPartOf\":{\"@id\":\"https:\/\/17aitech.com\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/17aitech.com\/?p=34453#primaryimage\"},\"image\":{\"@id\":\"https:\/\/17aitech.com\/?p=34453#primaryimage\"},\"thumbnailUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-ace45b8b0af65f751c1a14194401fe7b.png\",\"datePublished\":\"2024-12-06T14:56:23+00:00\",\"author\":{\"@id\":\"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d\"},\"breadcrumb\":{\"@id\":\"https:\/\/17aitech.com\/?p=34453#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/17aitech.com\/?p=34453\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/17aitech.com\/?p=34453#primaryimage\",\"url\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-ace45b8b0af65f751c1a14194401fe7b.png\",\"contentUrl\":\"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-ace45b8b0af65f751c1a14194401fe7b.png\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/17aitech.com\/?p=34453#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/17aitech.com\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR&#8217;24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b\"}]},{\"@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":"\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR'24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b - \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=34453","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/17aitech.com\/?p=34453","url":"https:\/\/17aitech.com\/?p=34453","name":"\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR'24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b - \u4e00\u8d77AI\u6280\u672f","isPartOf":{"@id":"https:\/\/17aitech.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/17aitech.com\/?p=34453#primaryimage"},"image":{"@id":"https:\/\/17aitech.com\/?p=34453#primaryimage"},"thumbnailUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-ace45b8b0af65f751c1a14194401fe7b.png","datePublished":"2024-12-06T14:56:23+00:00","author":{"@id":"https:\/\/17aitech.com\/#\/schema\/person\/60225458499e817ae0af73e67e440b9d"},"breadcrumb":{"@id":"https:\/\/17aitech.com\/?p=34453#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/17aitech.com\/?p=34453"]}]},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/17aitech.com\/?p=34453#primaryimage","url":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-ace45b8b0af65f751c1a14194401fe7b.png","contentUrl":"https:\/\/17aitech.com\/wp-content\/uploads\/2024\/07\/frc-ace45b8b0af65f751c1a14194401fe7b.png"},{"@type":"BreadcrumbList","@id":"https:\/\/17aitech.com\/?p=34453#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/17aitech.com\/"},{"@type":"ListItem","position":2,"name":"\u5f00\u6e90\u89c6\u9891\u7248GPT-4o\uff1f\u5feb\u901f\u8bb0\u5fc6\uff0c\u5b9e\u65f6\u95ee\u7b54\uff0c\u62ff\u4e0bCVPR&#8217;24\u957f\u89c6\u9891\u95ee\u7b54\u7ade\u8d5b\u51a0\u519b"}]},{"@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\/34453","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=34453"}],"version-history":[{"count":0,"href":"https:\/\/17aitech.com\/index.php?rest_route=\/wp\/v2\/posts\/34453\/revisions"}],"wp:attachment":[{"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=34453"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=34453"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/17aitech.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=34453"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}