https://www.dukehsu.com/tags/deepfake.htmlRecent content in Deepfake on Duke Hsu ｜ Learn • Build • Sharehttps://www.dukehsu.com/images/og-default.jpghttps://www.dukehsu.com/images/og-default.jpgHugoen-USFri, 06 Sep 2024 00:00:00 +0000https://www.dukehsu.com/posts/deeplivecaminstallnote.htmlFri, 06 Sep 2024 00:00:00 +0000https://www.dukehsu.com/posts/deeplivecaminstallnote.html<h1 id="deep-live-cam-笔记">Deep Live cam 笔记</h1> <p>项目地址： <a href="https://github.com/hacksider/Deep-Live-Cam">https://github.com/hacksider/Deep-Live-Cam</a></p> <p>在本篇笔记中不会涉及cuda,Python,Conda,Git,等相关的计算机知识。只有步骤记录和在使用部署时遇到的问题记录。</p> <p>以下操作都是在Windows操作系统下进行的，我的系统版本号为：</p> <p>Microsoft Windows 11 Pro<br> 10.0.22631 Build 22631</p> <p><img alt="image-20240906221442070.png" loading="lazy" src="https://www.dukehsu.com/posts/deeplivecaminstallnote/images/3366180621.png" title="image-20240906221442070.png"></p> <h2 id="先决条件">先决条件</h2> <ol> <li> <p>带有Nvidia 显卡的桌面电脑或者笔记本电脑</p> </li> <li> <p>稳定的网络环境</p> </li> <li> <p>DeepLiveCam所需的软件或编译工具</p> <ul> <li>Python 3.10.0</li> <li>CUDA</li> <li>cuDNN</li> <li>Git</li> <li>ffmpeg</li> <li>Conda-Py 3.10</li> <li>Microsoft C++ Build Tools</li> </ul> </li> <li> <p>DeepLiveCam项目本体</p> </li> <li> <p>模型下载</p> <ul> <li><a href="https://huggingface.co/hacksider/deep-live-cam/resolve/main/GFPGANv1.4.pth">GFPGANv1.4</a></li> <li><a href="https://huggingface.co/hacksider/deep-live-cam/resolve/main/inswapper_128_fp16.onnx">inswapper_128_fp16.onnx</a></li> </ul> </li> </ol> <h2 id="环境安装">环境安装</h2> <h3 id="python-3100">Python 3.10.0</h3> <p>下载地址： <a href="https://www.python.org/ftp/python/3.10.0/python-3.10.0-amd64.exe">https://www.python.org/ftp/python/3.10.0/python-3.10.0-amd64.exe</a></p> <p>像正常安装应用软件那样安装即可，记得勾选添加到系统变量</p> <h3 id="cuda">CUDA</h3> <p>下载地址： <a href="https://developer.nvidia.com/cuda-toolkit-archive">https://developer.nvidia.com/cuda-toolkit-archive</a></p> <p>安装完成之后在命令行运行 <code>nvidia-smi</code> 检查显卡支持的最大CUDA 版本</p> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-text" data-lang="text"><span class="line"><span class="cl">nvidia-smi</span></span></code></pre></div><p><img alt="image-20240906212930253.png" loading="lazy" src="https://www.dukehsu.com/posts/deeplivecaminstallnote/images/1102144062.png" title="image-20240906212930253.png"></p> <h3 id="cudnn">cuDNN</h3> <p>需要下载CUDA支持的版本</p> <p>下载地址： <a href="http://developer.nvidia.com/rdp/cudnn-archive">http://developer.nvidia.com/rdp/cudnn-archive</a></p> <p>安装完成之后在命令行运行 <code>nvcc -V</code> cuDNN的版本号</p> <p>如果提示命令不存在，请检查系统的环境变量和用户变量</p> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-text" data-lang="text"><span class="line"><span class="cl">nvcc -V</span></span></code></pre></div><p><img alt="image-20240906213133117.png" loading="lazy" src="https://www.dukehsu.com/posts/deeplivecaminstallnote/images/451047190.png" title="image-20240906213133117.png"></p> <h3 id="git">Git</h3> <p>下载地址： <a href="https://git-scm.com/downloads">https://git-scm.com/downloads</a></p> <p>安装完成之后在命令行运行 <code>git --version</code> 检查Git的版本号</p> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-text" data-lang="text"><span class="line"><span class="cl">git --version</span></span></code></pre></div><p>如果提示命令不存在，请检查系统的环境变量和用户变量</p> <p><img alt="image-20240906213419363.png" loading="lazy" src="https://www.dukehsu.com/posts/deeplivecaminstallnote/images/1811688303.png" title="image-20240906213419363.png"></p> <h3 id="conda">Conda</h3> <p>下载地址： <a href="https://docs.anaconda.com/miniconda/">https://docs.anaconda.com/miniconda/</a></p> <p>选择<code>Miniconda3-py310_24.7.1-0-Windows-x86_64.exe</code></p>