<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[m0_63257870的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/m0_63257870</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; m0_63257870]]></copyright><item><title><![CDATA[图片的维度进行转置：np.transpose()]]></title><link>https://blog.youkuaiyun.com/m0_63257870/article/details/142141911</link><guid>https://blog.youkuaiyun.com/m0_63257870/article/details/142141911</guid><author>m0_63257870</author><pubDate>Wed, 11 Sep 2024 15:42:46 +0800</pubDate><description><![CDATA[但在很多深度学习框架中（如PyTorch和TensorFlow），期望的输入维度是。函数对图片的维度进行转置。原始图片的维度通常是。]]></description><category></category></item><item><title><![CDATA[module ‘torch‘ has no attribute ‘gesv‘]]></title><link>https://blog.youkuaiyun.com/m0_63257870/article/details/141569008</link><guid>https://blog.youkuaiyun.com/m0_63257870/article/details/141569008</guid><author>m0_63257870</author><pubDate>Mon, 26 Aug 2024 17:13:11 +0800</pubDate><description><![CDATA[【代码】module ‘torch‘ has no attribute ‘gesv‘]]></description><category></category></item><item><title><![CDATA[切换CUDA版本]]></title><link>https://blog.youkuaiyun.com/m0_63257870/article/details/141563635</link><guid>https://blog.youkuaiyun.com/m0_63257870/article/details/141563635</guid><author>m0_63257870</author><pubDate>Mon, 26 Aug 2024 14:55:24 +0800</pubDate><description><![CDATA[CUDA版本切换]]></description><category></category></item><item><title><![CDATA[RecDiffusion: Rectangling for Image Stitching with Diffusion Models]]></title><link>https://blog.youkuaiyun.com/m0_63257870/article/details/138648060</link><guid>https://blog.youkuaiyun.com/m0_63257870/article/details/138648060</guid><author>m0_63257870</author><pubDate>Tue, 14 May 2024 14:56:52 +0800</pubDate><description><![CDATA[一种新颖的基于扩散的学习框架 RecDiffusion。该框架结合运动扩散模型（MDM）生成运动场，有效地从拼接图像的不规则边界过渡到几何校正的中间体。随后是内容扩散模型（CDM），用于图像细节细化。值得注意的是，在 CDM 的每次迭代中，我们的采样过程都会利用加权地图来识别需要校正的区域。我们的 RecDiffusion 可确保几何精度和整体视觉效果，在公共基准上进行评估时，其定量和定性指标均超过了以往所有方法。]]></description><category></category></item></channel></rss>