{"id":23395,"date":"2022-08-05T07:41:46","date_gmt":"2022-08-04T23:41:46","guid":{"rendered":"http:\/\/www.gocpmall.com\/?p=23395"},"modified":"2022-08-05T07:41:46","modified_gmt":"2022-08-04T23:41:46","slug":"%e8%a7%86%e9%a2%91%e4%b9%90%e9%ab%98%e4%b8%8d%e4%bc%9a%e6%8b%bc%e4%ba%86%ef%bc%9f%e8%bf%99%e6%ac%beai%e8%83%bd%e5%b8%ae%e4%bd%a0%e6%a0%b9%e6%8d%ae%e6%89%8b%e5%86%8c%e6%8b%bc%e6%90%ad%e5%87%ba","status":"publish","type":"post","link":"https:\/\/www.gocpmall.com\/?p=23395","title":{"rendered":"[\u89c6\u9891]\u4e50\u9ad8\u4e0d\u4f1a\u62fc\u4e86\uff1f\u8fd9\u6b3eAI\u80fd\u5e2e\u4f60\u6839\u636e\u624b\u518c\u62fc\u642d\u51fa\u6210\u54c1"},"content":{"rendered":"

\"news
访问： 阿里云服务器精选特惠：1核1G云服务器低至0.9元\/月 该框架叫做 Manual-to-Executable-Plan Network（简称 MEPNet），目前已经在多款计算机生成的乐高套装、真实的乐高套装指南和 Minecraft 风格的立体建筑上成功测试，科研人员表示其性能已经超过现有其他版本。 对于人工智能来说，理解 2D 指导手册并不容易。研究人员表示，视觉指令有几个关键问题，就像乐高套装一样，完全由图像组成：识别 2D 和 3D 对象之间的对应关系，以及处理许多基本部件都非常困难。 研究人员表示在，任何复杂的乐高套装都是建立在基础的乐高砖块上。研究人员表示这增加了机器对乐高手册的理解难度，它需要推断由可见图元组成的不可见物体的 3D 姿势。 研究人员表示，现有的将手动步骤解析为机器可执行计划的方法主要包括两种形式：基于搜索的方法，简单准确但计算成本高；以及基于学习的模型，这些模型速度很快，但不太擅长处理看不见的 3D 形状。 研究人员说，MEPNet 结合了两者。研究人员写道，从组件的 3D 模型、乐高集的当前状态和 2D 手动图像开始，MEPNet “预测每个组件的一组 2D 关键点和掩码”。 完成后，2D 关键点“通过找到基本形状和新组件之间的可能连接，反向投影到 3D”。该团队写道，这种组合“保持了基于学习的模型的效率，并更好地推广到看不见的 3D 组件”。在论文中，研究人员表示，他们的目标是创造帮助人们组装复杂物体的机器，他们的应用列表中包括家具、乐高积木和像素世界。
阅读更多（Read More）<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

访问： 阿里云&#2638 […]<\/p>\n","protected":false},"author":2,"featured_media":23396,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[24],"tags":[],"class_list":["post-23395","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-localnews"],"_links":{"self":[{"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=\/wp\/v2\/posts\/23395","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=23395"}],"version-history":[{"count":0,"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=\/wp\/v2\/posts\/23395\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=\/wp\/v2\/media\/23396"}],"wp:attachment":[{"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=23395"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=23395"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.gocpmall.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=23395"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}