{"id":44995,"date":"2025-10-21T12:28:15","date_gmt":"2025-10-21T04:28:15","guid":{"rendered":"https:\/\/www.1ai.net\/?p=44995"},"modified":"2025-10-21T12:28:15","modified_gmt":"2025-10-21T04:28:15","slug":"%e5%8d%95%e5%8d%a1%e6%97%a5%e5%a4%84%e7%90%86-20-%e4%b8%87%e9%a1%b5%e6%96%87%e6%a1%a3%ef%bc%8cdeepseek-ocr-%e5%bc%80%e6%ba%90%e4%b8%8a%e7%ba%bf","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/44995.html","title":{"rendered":"Single card day processing 200,000 pages of documents, DeepSeek-OCR open-source"},"content":{"rendered":"<p>10 \u6708 21 \u65e5\u6d88\u606f\uff0c\u636e\u300cAl\u5bd2\u6b66\u7eaa\u300d\u62a5\u9053\uff0c<a href=\"https:\/\/www.1ai.net\/en\/tag\/deepseek\" title=\"\u3010\u67e5\u770b\u542b\u6709[DeepSeek]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">DeepSeek<\/a> \u56e2\u961f\u8fd1\u65e5\u53d1\u5e03\u5168\u65b0\u7814\u7a76\u6210\u679c DeepSeek-<a href=\"https:\/\/www.1ai.net\/en\/tag\/ocr\" title=\"\u3010\u67e5\u770b\u542b\u6709[OCR]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">OCR<\/a>\uff0c\u63d0\u51fa\u300c\u4e0a\u4e0b\u6587\u5149\u5b66\u538b\u7f29\u300d\u65b9\u6cd5\uff0c\u4e3a\u5927\u6a21\u578b\u957f\u6587\u672c\u5904\u7406\u5e26\u6765\u7a81\u7834\u6027\u601d\u8def\u3002<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-44996\" title=\"b9557285j00t4gsd5004rd000u000hnm\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/10\/b9557285j00t4gsd5004rd000u000hnm.jpg\" alt=\"b9557285j00t4gsd5004rd000u000hnm\" width=\"1080\" height=\"635\" \/><\/p>\n<p>\u7814\u7a76\u663e\u793a\uff0c\u901a\u8fc7\u5c06\u957f\u6587\u672c\u6e32\u67d3\u4e3a\u56fe\u50cf\uff0c\u518d\u8f6c\u5316\u4e3a\u89c6\u89c9 token\uff0c\u80fd\u591f\u5728\u4fdd\u6301\u9ad8\u7cbe\u5ea6\u7684\u540c\u65f6\u663e\u8457\u964d\u4f4e\u8ba1\u7b97\u6210\u672c\u3002<\/p>\n<p>\u5b9e\u9a8c\u6570\u636e\u663e\u793a\uff0c\u5728\u5c0f\u4e8e 10 \u500d\u538b\u7f29\u7387\u4e0b\uff0cOCR \u89e3\u7801\u51c6\u786e\u7387\u9ad8\u8fbe 97%\uff1b\u5373\u4fbf\u5728 20 \u500d\u8d85\u9ad8\u538b\u7f29\u7387\u4e0b\uff0c\u51c6\u786e\u7387\u4ecd\u80fd\u7ef4\u6301\u5728\u7ea6 60%\u3002\u5728\u6743\u5a01\u6587\u6863\u89e3\u6790\u57fa\u51c6 OmniDocBench \u4e0a\uff0c\u8be5\u6a21\u578b\u4ee5\u66f4\u5c11\u7684\u89c6\u89c9 token \u8d85\u8d8a\u4e86\u591a\u9879\u4e3b\u6d41 SOTA \u6a21\u578b\u3002<\/p>\n<p>\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\uff0c\u5355\u53f0 A100-40G GPU \u6bcf\u5929\u53ef\u5904\u7406\u8d85\u8fc7 20 \u4e07\u9875\u6587\u6863\uff0c\u4e3a\u5927\u6a21\u578b\u8bad\u7ec3\u63d0\u4f9b\u6d77\u91cf\u6570\u636e\u652f\u6301\u3002<\/p>\n<p>\u76ee\u524d\uff0c\u76f8\u5173\u4ee3\u7801\u4e0e\u6a21\u578b\u6743\u91cd\u5df2\u5728 GitHub \u4e0e Hugging Face \u5e73\u53f0<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e5%bc%80%e6%ba%90\" title=\"\u3010\u67e5\u770b\u542b\u6709[\u5f00\u6e90]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">\u5f00\u6e90<\/a>\u3002<\/p>\n<p>\ud83d\udcbb GitHub\uff1ahttps:\/\/github.com\/deepseek-ai\/DeepSeek-OCR\/<\/p>\n<p>\ud83e\udd17 Hugging Face\uff1ahttps:\/\/huggingface.co\/deepseek-ai\/DeepSeek-OCR<\/p>\n","protected":false},"excerpt":{"rendered":"<p>According to news from October 21, the DeepSeek team has recently released a new study, DeepSeek-OCR, which proposes a \"text-based optical compression\" approach that provides groundbreaking thinking for long text processing for large models. Research shows that by rendering long text into images and then turning to visual token, it is possible to significantly reduce the calculation costs while maintaining high accuracy. Experimental data show that the OCR decoded accuracy rate was as high as 971 TP3T at a rate of less than 10 times; even at a rate of 20 times higher, the accuracy rate remained at about 601 TP3T. On the authoritative document parsing baseline OmniDocBench, the model transcends many mainstream SOTA with less visual token<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[146],"tags":[3606,4860,219],"collection":[],"class_list":["post-44995","post","type-post","status-publish","format-standard","hentry","category-news","tag-deepseek","tag-ocr","tag-219"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/44995","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=44995"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/44995\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=44995"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=44995"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=44995"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=44995"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}