{"id":18219,"date":"2025-01-13T20:48:16","date_gmt":"2025-01-13T12:48:16","guid":{"rendered":"https:\/\/www.aisharenet.com\/?p=18219"},"modified":"2025-01-13T20:48:16","modified_gmt":"2025-01-13T12:48:16","slug":"star","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/en\/star\/","title":{"rendered":"STAR\uff1a\u7a7a\u95f4\u65f6\u5e8f\u589e\u5f3aAI\u6a21\u578b\uff0c\u63d0\u5347\u89c6\u9891\u5206\u8fa8\u7387\u6e05\u6670\u5ea6"},"content":{"rendered":"<p>STAR (Spatial-Temporal Augmentation with Text-to-Video Models) \u662f\u7531\u5357\u4eac\u5927\u5b66\u3001\u5b57\u8282\u8df3\u52a8\u548c\u897f\u5357\u5927\u5b66\u8054\u5408\u5f00\u53d1\u7684\u521b\u65b0\u578b\u89c6\u9891\u8d85\u5206\u8fa8\u7387\u6846\u67b6\u3002\u8be5\u9879\u76ee\u81f4\u529b\u4e8e\u89e3\u51b3\u771f\u5b9e\u4e16\u754c\u89c6\u9891\u8d85\u5206\u8fa8\u7387\u5904\u7406\u4e2d\u7684\u5173\u952e\u95ee\u9898\uff0c\u901a\u8fc7\u7ed3\u5408\u6587\u672c\u5230\u89c6\u9891(T2V)\u6269\u6563\u6a21\u578b\u7684\u5148\u9a8c\u77e5\u8bc6\uff0c\u5b9e\u73b0\u4e86\u89c6\u9891\u753b\u9762\u7684\u9ad8\u8d28\u91cf\u589e\u5f3a\u3002STAR\u6a21\u578b\u7684\u7a81\u51fa\u7279\u70b9\u5728\u4e8e\u5176\u80fd\u591f\u540c\u65f6\u4fdd\u6301\u7a7a\u95f4\u7ec6\u8282\u7684\u771f\u5b9e\u6027\u548c\u65f6\u5e8f\u4e00\u81f4\u6027\uff0c\u8fd9\u5728\u4f20\u7edf\u7684GAN based\u65b9\u6cd5\u4e2d\u5f80\u5f80\u96be\u4ee5\u517c\u987e\u3002\u9879\u76ee\u63d0\u4f9b\u4e86\u4e24\u4e2a\u7248\u672c\u7684\u5b9e\u73b0\uff1a\u57fa\u4e8eI2VGen-XL\u7684\u8f7b\u91cf\u7ea7\u548c\u91cd\u5ea6\u964d\u8d28\u5904\u7406\u6a21\u578b\uff0c\u4ee5\u53ca\u57fa\u4e8eCogVideoX-5B\u7684\u91cd\u5ea6\u964d\u8d28\u5904\u7406\u6a21\u578b\uff0c\u80fd\u591f\u9002\u5e94\u4e0d\u540c\u573a\u666f\u4e0b\u7684\u89c6\u9891\u589e\u5f3a\u9700\u6c42\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"aligncenter size-full wp-image-18221\" title=\"STAR\uff1a\u7a7a\u95f4\u65f6\u5e8f\u589e\u5f3aAI\u6a21\u578b\uff0c\u63d0\u5347\u89c6\u9891\u5206\u8fa8\u7387\u6e05\u6670\u5ea6-1\" src=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/01\/5354bc28f325a5d.png\" alt=\"STAR\uff1a\u7a7a\u95f4\u65f6\u5e8f\u589e\u5f3aAI\u6a21\u578b\uff0c\u63d0\u5347\u89c6\u9891\u5206\u8fa8\u7387\u6e05\u6670\u5ea6-1\" width=\"1200\" height=\"484\" srcset=\"https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/01\/5354bc28f325a5d.png 1200w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/01\/5354bc28f325a5d-300x121.png 300w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/01\/5354bc28f325a5d-1024x413.png 1024w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/01\/5354bc28f325a5d-768x310.png 768w, https:\/\/www.kdjingpai.com\/wp-content\/uploads\/2025\/01\/5354bc28f325a5d-18x7.png 18w\" sizes=\"auto, (max-width: 1200px) 100vw, 1200px\" \/><\/p>\n<p>&nbsp;<\/p>\n<h2>\u529f\u80fd\u5217\u8868<\/h2>\n<ul>\n<li>\u652f\u6301\u591a\u79cd\u7c7b\u578b\u89c6\u9891\u964d\u8d28\u5904\u7406\uff08\u8f7b\u5ea6\u548c\u91cd\u5ea6\uff09\u7684\u8d85\u5206\u8fa8\u7387\u91cd\u5efa<\/li>\n<li>\u81ea\u52a8\u5316\u63d0\u793a\u8bcd\u751f\u6210\u529f\u80fd\uff0c\u652f\u6301\u4f7f\u7528Pllava\u7b49\u5de5\u5177\u751f\u6210\u89c6\u9891\u63cf\u8ff0<\/li>\n<li>\u63d0\u4f9b\u5728\u7ebfDemo\u6f14\u793a\u5e73\u53f0\uff08HuggingFace Spaces\uff09<\/li>\n<li>\u652f\u6301720&#215;480\u5206\u8fa8\u7387\u7684\u89c6\u9891\u8f93\u5165\u5904\u7406<\/li>\n<li>\u63d0\u4f9b\u5b8c\u6574\u7684\u63a8\u7406\u4ee3\u7801\u548c\u9884\u8bad\u7ec3\u6a21\u578b<\/li>\n<li>\u96c6\u6210\u672c\u5730\u4fe1\u606f\u589e\u5f3a\u6a21\u5757(LIEM)\uff0c\u63d0\u5347\u753b\u9762\u7ec6\u8282\u91cd\u5efa\u8d28\u91cf<\/li>\n<li>\u652f\u6301\u6279\u91cf\u89c6\u9891\u5904\u7406<\/li>\n<li>\u63d0\u4f9b\u7075\u6d3b\u7684\u6a21\u578b\u6743\u91cd\u9009\u62e9<\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<h2>\u4f7f\u7528\u5e2e\u52a9<\/h2>\n<h3>1. \u73af\u5883\u914d\u7f6e<\/h3>\n<p>\u9996\u5148\u9700\u8981\u914d\u7f6e\u8fd0\u884c\u73af\u5883\uff0c\u5177\u4f53\u6b65\u9aa4\u5982\u4e0b\uff1a<\/p>\n<ol>\n<li>\u514b\u9686\u4ee3\u7801\u4ed3\u5e93\uff1a<\/li>\n<\/ol>\n<pre><code>git clone https:\/\/github.com\/NJU-PCALab\/STAR.git\r\ncd STAR\r\n<\/code><\/pre>\n<ol>\n<li>\u521b\u5efa\u5e76\u6fc0\u6d3bconda\u73af\u5883\uff1a<\/li>\n<\/ol>\n<pre><code>conda create -n star python=3.10\r\nconda activate star\r\npip install -r requirements.txt\r\nsudo apt-get update &amp;&amp; apt-get install ffmpeg libsm6 libxext6 -y\r\n<\/code><\/pre>\n<h3>2. \u6a21\u578b\u9009\u62e9\u548c\u4e0b\u8f7d<\/h3>\n<p>STAR\u63d0\u4f9b\u4e86\u4e24\u79cd\u7248\u672c\u7684\u6a21\u578b\uff1a<\/p>\n<ul>\n<li>I2VGen-XL based\u6a21\u578b\uff1a\n<ul>\n<li>light_deg.pt\uff1a\u9002\u7528\u4e8e\u8f7b\u5ea6\u964d\u8d28\u89c6\u9891\u5904\u7406<\/li>\n<li>heavy_deg.pt\uff1a\u9002\u7528\u4e8e\u91cd\u5ea6\u964d\u8d28\u89c6\u9891\u5904\u7406<\/li>\n<\/ul>\n<\/li>\n<li>CogVideoX-5B based\u6a21\u578b\uff1a\n<ul>\n<li>\u4e13\u95e8\u7528\u4e8e\u5904\u7406\u91cd\u5ea6\u964d\u8d28\u89c6\u9891<\/li>\n<li>\u4ec5\u652f\u6301720&#215;480\u5206\u8fa8\u7387\u8f93\u5165<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p>\u4eceHuggingFace\u4e0b\u8f7d\u76f8\u5e94\u6a21\u578b\u6743\u91cd\u5e76\u653e\u7f6e\u5728<code>pretrained_weight\/<\/code>\u76ee\u5f55\u4e0b\u3002<\/p>\n<h3>3. \u89c6\u9891\u5904\u7406\u6d41\u7a0b<\/h3>\n<ol>\n<li>\u51c6\u5907\u6d4b\u8bd5\u6570\u636e\uff1a\n<ul>\n<li>\u5c06\u5f85\u5904\u7406\u89c6\u9891\u653e\u5165<code>input\/video\/<\/code>\u76ee\u5f55<\/li>\n<li>\u63d0\u793a\u8bcd\u51c6\u5907\uff08\u4e09\u79cd\u9009\u62e9\uff09\uff1a\n<ul>\n<li>\u65e0\u63d0\u793a\u8bcd<\/li>\n<li>\u4f7f\u7528Pllava\u81ea\u52a8\u751f\u6210<\/li>\n<li>\u624b\u52a8\u7f16\u5199\u89c6\u9891\u63cf\u8ff0<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/li>\n<li>\u914d\u7f6e\u5904\u7406\u53c2\u6570\uff1a\n<ul>\n<li>\u4fee\u6539<code>video_super_resolution\/scripts\/inference_sr.sh<\/code>\u4e2d\u7684\u8def\u5f84\u914d\u7f6e\uff1a\n<ul>\n<li>video_folder_path\uff1a\u8f93\u5165\u89c6\u9891\u8def\u5f84<\/li>\n<li>txt_file_path\uff1a\u63d0\u793a\u8bcd\u6587\u4ef6\u8def\u5f84<\/li>\n<li>model_path\uff1a\u6a21\u578b\u6743\u91cd\u8def\u5f84<\/li>\n<li>save_dir\uff1a\u8f93\u51fa\u4fdd\u5b58\u8def\u5f84<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/li>\n<li>\u542f\u52a8\u63a8\u7406\uff1a<\/li>\n<\/ol>\n<pre><code>bash video_super_resolution\/scripts\/inference_sr.sh\r\n<\/code><\/pre>\n<p>\u6ce8\u610f\uff1a\u5982\u9047\u5230\u5185\u5b58\u6ea2\u51fa(OOM)\u95ee\u9898\uff0c\u53ef\u4ee5\u5728<code>inference_sr.sh<\/code>\u4e2d\u8c03\u5c0f<code>frame_length<\/code>\u53c2\u6570\u3002<\/p>\n<h3>4. CogVideoX-5B\u6a21\u578b\u7279\u6b8a\u914d\u7f6e<\/h3>\n<p>\u5982\u679c\u4f7f\u7528CogVideoX-5B\u6a21\u578b\uff0c\u9700\u8981\u989d\u5916\u6b65\u9aa4\uff1a<\/p>\n<ol>\n<li>\u521b\u5efa\u4e13\u7528\u73af\u5883\uff1a<\/li>\n<\/ol>\n<pre><code>conda create -n star_cog python=3.10\r\nconda activate star_cog\r\ncd cogvideox-based\/sat\r\npip install -r requirements.txt\r\n<\/code><\/pre>\n<ol>\n<li>\u4e0b\u8f7d\u989d\u5916\u4f9d\u8d56\uff1a<\/li>\n<\/ol>\n<ul>\n<li>\u9700\u8981\u4e0b\u8f7dVAE\u548cT5 Encoder<\/li>\n<li>\u66f4\u65b0<code>cogvideox-based\/sat\/configs\/cogvideox_5b\/cogvideox_5b_infer_sr.yaml<\/code>\u4e2d\u7684\u8def\u5f84\u914d\u7f6e<\/li>\n<li>\u66ff\u6362transformer.py\u6587\u4ef6<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>STAR (Spatial-Temporal Augmentation with Text-to-Video Models) \u662f\u7531\u5357\u4eac\u5927\u5b66\u3001\u5b57\u8282\u8df3\u52a8\u548c\u897f\u5357\u5927\u5b66\u8054\u5408\u5f00\u53d1\u7684\u521b\u65b0\u578b\u89c6\u9891\u8d85\u5206\u8fa8\u7387\u6846\u67b6\u3002\u8be5\u9879\u76ee\u81f4\u529b\u4e8e\u89e3\u51b3\u771f\u5b9e\u4e16\u754c\u89c6\u9891\u8d85\u5206\u8fa8\u7387\u5904\u7406\u4e2d\u7684\u5173\u952e&#8230;<\/p>\n","protected":false},"author":1,"featured_media":61619,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[218,230],"class_list":["post-18219","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-tool","tag-aituxiangfangda","tag-aikaiyuanxiangmu"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/18219","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/comments?post=18219"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/posts\/18219\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media\/61619"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/media?parent=18219"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/categories?post=18219"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/en\/wp-json\/wp\/v2\/tags?post=18219"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}