
{"id":13795,"date":"2026-10-23T14:17:43","date_gmt":"2026-10-23T06:17:43","guid":{"rendered":"https:\/\/infernews.com\/blog\/textriro\/"},"modified":"2026-10-23T14:24:08","modified_gmt":"2026-10-23T06:24:08","slug":"textriro","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/textriro\/","title":{"rendered":"TextRIRO\uff1a\u4fee\u6539\u6587\u5b57\u5167\u5bb9\u4e26\u4fdd\u7559\u539f\u59cb\u98a8\u683c"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/10\/TextRIRO.jpg\" alt=\"TextRIRO training, inference, STR-pretrained text encoder, and TextRIRO-3M construction pipeline\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u62db\u724c\u6216\u5305\u88dd\u4e0a\u7684\u5b57\u63db\u4e86\uff0c\u4ecd\u5e0c\u671b\u4fdd\u7559\u539f\u6709\u5916\u89c0\uff1bTextRIRO \u4ee5\u689d\u4ef6\u64f4\u6563\u6a21\u578b\u8655\u7406\u5834\u666f\u6587\u5b57\u7de8\u8f2f\uff08Scene Text Editing\uff0cSTE\uff09\uff0c\u8f38\u5165\u88c1\u5207\u6587\u5b57\u5716\u53ca\u66ff\u63db\u5167\u5bb9\uff0c\u8f38\u51fa\u6539\u5b57\u5f8c\u7684\u5716\u50cf\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b83\u4ee5 Shuffle&amp;Crop \u6253\u4e82\u771f\u5be6\u5f71\u50cf\u7684\u6587\u5b57\u8a9e\u610f\uff0c\u540c\u6642\u4fdd\u7559\u98a8\u683c\u7dda\u7d22\uff0c\u8b93\u6bcf\u5f35\u5716\u5f62\u6210\u81ea\u8eab\u7684\u914d\u5c0d\u76e3\u7763\uff0c\u6bcb\u9808\u5408\u6210\u7de8\u8f2f\u524d\u5f8c\u914d\u5c0d\u6216\u5b57\u5143\u5206\u5272\u3002\u591a\u5c3a\u5ea6\u98a8\u683c\u7de8\u78bc\u5668\u914d\u5408\u4ee5\u5834\u666f\u6587\u5b57\u8fa8\u8b58\uff08Scene Text Recognition\uff0cSTR\uff09\u9810\u8a13\u7df4\u7684\u6587\u5b57\u7de8\u78bc\u5668\uff0c\u5206\u5225\u63d0\u4f9b\u5916\u89c0\u8207\u5b57\u5143\u689d\u4ef6\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u300cReal-only\u300d\u53ea\u6db5\u84cb\u7de8\u8f2f\u76e3\u7763\u53ca\u6587\u5b57\u7de8\u78bc\u5668\u9810\u8a13\u7df4\u3002\u751f\u6210\u9aa8\u5e79\u4ecd\u63a1\u7528 Stable Diffusion\uff08SD\uff091.5 \u7684 U-Net \u8207 scheduler\uff0c\u4ee5\u53ca SD 2.1 \u7684\u8b8a\u5206\u81ea\u7de8\u78bc\u5668\uff08Variational Autoencoder\uff0cVAE\uff09\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u53ef\u900f\u904e <a href=\"https:\/\/huggingface.co\/spaces\/Yesianrohn\/TextRIRO-Demo\" target=\"_blank\" rel=\"noopener\">Hugging Face Demo<\/a> \u8a66\u6539\u6587\u5b57\u3002<\/li>\n\n\n\n<li>\u672c\u5730\u63a8\u7406\u9700 CUDA\u3001\u6a21\u578b\u6b0a\u91cd\u53ca\u76f8\u95dc\u7d44\u4ef6\uff1b\u5132\u5b58\u5eab\u5217\u51fa Python 3.8\uff0fPyTorch 1.13 \u74b0\u5883\u3002<\/li>\n\n\n\n<li>\u751f\u6210\u89e3\u50cf\u5ea6\u70ba 64 \u00d7 256\uff0c\u518d\u9084\u539f\u4f86\u6e90\u5c3a\u5bf8\uff0c\u7d30\u7bc0\u4ecd\u53d7\u9650\u5236\u3002<\/li>\n\n\n\n<li>RIROCommon \u6e2c\u8a66\u7b49\u9577\u53ca\u8b8a\u9577\u6539\u5b57\uff1b\u6240\u9644\u8cc7\u8a0a\u672a\u63d0\u4f9b\u5206\u6578\uff0c\u4e0d\u80fd\u5224\u65b7\u9818\u5148\u5e45\u5ea6\u3002<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u6587\u5b57\u8fa8\u8b58\u7814\u7a76\u5718\u968a\u4ea6\u53ef\u5229\u7528 TextRIRO-3M \u7684\u4e09\u767e\u842c\u5f35\u9577\u6587\u5b57\u5f71\u50cf\u3002\u9019\u6279\u8cc7\u6599\u7531\u7be9\u9078\u5f8c\u7684\u7de8\u8f2f\u5716\u8207\u539f\u5716\u62fc\u63a5\u800c\u6210\uff0c\u5c6c\u5408\u6210\u8cc7\u6599\uff1b\u5b8c\u6574\u7167\u7247\u6539\u5b57\u6d41\u7a0b\uff0c\u4ecd\u9700\u53e6\u884c\u8655\u7406\u6587\u5b57\u5340\u57df\u88c1\u5207\u53ca\u56de\u8cbc\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/github.com\/YesianRohn\/TextRIRO\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a> \u00b7 <a href=\"https:\/\/huggingface.co\/Yesianrohn\/TextRIRO\" rel=\"noopener noreferrer\"><strong>\u6a21\u578b<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u8b93\u5716\u7247\u4e2d\u7684\u6587\u5b57\u63db\u5167\u5bb9\u3001\u7559\u98a8\u683c\uff0c\u4f46\u8655\u7406\u7bc4\u570d\u9650\u65bc\u88c1\u5207\u6587\u5b57\u5340\u57df\uff0c\u9805\u76ee\u66ab\u6642\u53ea\u662f\u4f5c\u70ba\u7814\u7a76\uff0c\u56e0\u6b64\u751f\u6210\u89e3\u50cf\u5ea6\u4ea6\u6709\u9650\u5236\u3002\u6211\u4ecd\u70ba\u9019\u7814\u7a76\u6709\u52a9 Diffusion Model \u767c\u5c55\u66f4\u591a\u7684\u61c9\u7528<\/p>\n","protected":false},"author":8,"featured_media":13794,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"wpai_generated_summary":"","wpai_meta_description":"","footnotes":""},"categories":[133,76,127,179,33,30,152,150,199],"tags":[],"class_list":["post-13795","post","type-post","status-publish","format-standard","hentry","category-133","category-76","category-127","category-nvidia","category-stablediffusion","category-image","category-python","category-150","category-dataset-"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13795","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=13795"}],"version-history":[{"count":3,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13795\/revisions"}],"predecessor-version":[{"id":13800,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/13795\/revisions\/13800"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/13794"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=13795"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=13795"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=13795"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}