[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-70914":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":16,"stars30d":17,"stars90d":15,"forks30d":15,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":22,"defaultBranch":23,"hasWiki":21,"hasPages":22,"topics":24,"createdAt":9,"pushedAt":9,"updatedAt":25,"readmeContent":26,"aiSummary":27,"trendingCount":15,"starSnapshotCount":15,"syncStatus":13,"lastSyncTime":28,"discoverSource":29},70914,"conv_arithmetic","vdumoulin\u002Fconv_arithmetic","vdumoulin","A technical report on convolution arithmetic in the context of deep learning",null,"TeX",14646,2314,2,6,0,1,5,3,45,"MIT License",true,false,"master",[],"2026-06-12 02:02:45","# Convolution arithmetic\n\nA technical report on convolution arithmetic in the context of deep learning.\n\nThe code and the images of this tutorial are free to use as regulated by the \nlicence and subject to proper attribution:\n\n* \\[1\\] Vincent Dumoulin, Francesco Visin - [A guide to convolution arithmetic\n  for deep learning](https:\u002F\u002Farxiv.org\u002Fabs\u002F1603.07285)\n  ([BibTeX](https:\u002F\u002Fgist.github.com\u002Ffvisin\u002F165ca9935392fa9600a6c94664a01214))\n\n## Convolution animations\n\n_N.B.: Blue maps are inputs, and cyan maps are outputs._\n\n\u003Ctable style=\"width:100%; table-layout:fixed;\">\n  \u003Ctr>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fno_padding_no_strides.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Farbitrary_padding_no_strides.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fsame_padding_no_strides.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Ffull_padding_no_strides.gif\">\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>No padding, no strides\u003C\u002Ftd>\n    \u003Ctd>Arbitrary padding, no strides\u003C\u002Ftd>\n    \u003Ctd>Half padding, no strides\u003C\u002Ftd>\n    \u003Ctd>Full padding, no strides\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fno_padding_strides.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fpadding_strides.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fpadding_strides_odd.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>No padding, strides\u003C\u002Ftd>\n    \u003Ctd>Padding, strides\u003C\u002Ftd>\n    \u003Ctd>Padding, strides (odd)\u003C\u002Ftd>\n    \u003Ctd>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## Transposed convolution animations\n\n_N.B.: Blue maps are inputs, and cyan maps are outputs._\n\n\u003Ctable style=\"width:100%; table-layout:fixed;\">\n  \u003Ctr>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fno_padding_no_strides_transposed.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Farbitrary_padding_no_strides_transposed.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fsame_padding_no_strides_transposed.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Ffull_padding_no_strides_transposed.gif\">\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>No padding, no strides, transposed\u003C\u002Ftd>\n    \u003Ctd>Arbitrary padding, no strides, transposed\u003C\u002Ftd>\n    \u003Ctd>Half padding, no strides, transposed\u003C\u002Ftd>\n    \u003Ctd>Full padding, no strides, transposed\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fno_padding_strides_transposed.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fpadding_strides_transposed.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fpadding_strides_odd_transposed.gif\">\u003C\u002Ftd>\n    \u003Ctd>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>No padding, strides, transposed\u003C\u002Ftd>\n    \u003Ctd>Padding, strides, transposed\u003C\u002Ftd>\n    \u003Ctd>Padding, strides, transposed (odd)\u003C\u002Ftd>\n    \u003Ctd>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## Dilated convolution animations\n\n_N.B.: Blue maps are inputs, and cyan maps are outputs._\n\n\u003Ctable style=\"width:25%\"; table-layout:fixed;>\n  \u003Ctr>\n    \u003Ctd>\u003Cimg width=\"150px\" src=\"gif\u002Fdilation.gif\">\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>No padding, no stride, dilation\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## Generating the Makefile\n\nFrom the repository's root directory:\n\n``` bash\n$ .\u002Fbin\u002Fgenerate_makefile\n```\n## Generating the animations\n\nFrom the repository's root directory:\n\n``` bash\n$ make all_animations\n```\n\nThe animations will be output to the `gif` directory. Individual animation steps\nwill be output in PDF format to the `pdf` directory and in PNG format to the\n`png` directory.\n\n## Compiling the document\n\nFrom the repository's root directory:\n\n``` bash\n$ make\n```\n","该项目是一份关于深度学习中卷积算术的技术报告。它通过详细的数学推导和动画演示了不同类型的卷积操作，包括普通卷积、转置卷积以及空洞卷积等，并展示了这些操作在无填充\u002F有填充、无步幅\u002F有步幅情况下的具体表现形式。项目使用TeX语言编写，提供了丰富的可视化资源来帮助理解复杂的卷积概念。非常适合于正在学习或研究深度神经网络特别是计算机视觉领域的研究人员、学生及开发者参考使用。","2026-06-11 03:34:54","high_star"]