weixin_44109545的博客

前言：以下仅为个人在学习过程中的记录和总结，有问题欢迎友善讨论。仅以4D input为例（对应二维图片）torch.nn.functional.grid_sample(input, grid,mode='bilinear', padding_mode='zeros', align_corners=None)Args:input (N,C,Hin,Win) (N,C,H_{in}, W_{in})(N,C,Hin​,Win​)N：batch sizeC：feature dimen.

pytorch3d_installationRequirmentsRequirmentscore libiraryLinux or macOS or WindowsPython 3.6, 3.7 or 3.8PyTorch 1.4, 1.5.0, 1.5.1, 1.6.0, 1.7.0, or 1.7.1.torchvision that matches the PyTorch installation. You can install them together as explained

代码学习_KPConv_models.blocksclass KPConv(nn.Module)二级目录三级目录class KPConv(nn.Module)初始化赋值创建kernel points: K_points_numpy # Initialize weights [n_kpoints, in_fdim, out_fdim] self.weights = Parameter(torch.zeros((self.K, in_channels, out_cha

代码学习_KPConv_kernels.kernel_pointsload_kernels()kernel_point_optimization_debug()load_kernels()kernel_point_optimization_debug() """ Creation of kernel point via optimization of potentials. Args: radius: Radius of the kernels, 1

torch.utils.data.DataLoader作用：用来将数据包装为Dataset类，然后传入DataLoader中，再使用DataLoader这个类来更加快捷的对数据进行操作class DataLoader(object): r""" Data loader. Combines a dataset and a sampler, and provides single- or multi-process iterators over the dataset.

torch.nn.Conv1d Args: in_channels (int): Number of channels in the input image out_channels (int): Number of channels produced by the convolution kernel_size (int or tuple): Size of the convolving kernel stride (int or tupl

遇到问题：列表中的索引为一个列表记录代入具体值试验的结果example_1points = torch.rand((5, 4, 3))b_idx = torch.randint(4, (5, 2))idx = torch.randint(4, (5, 2))new = points[b_idx, idx, :]points: >>>tensor([[[5.9856e-01, 8.5678e-01, 9.8524e-01], [7.0864e-01.

记录_nn.functional.one_hot(x, num_classes)作用：将x的元素转化为0/1标签值Return.shape: (x.size, num_classes)num_classes 为目标类别数，需要>=实际类别数（输入x的元素值的类别）a = torch.randint(0, 24, (3, 4, 5)) >>>tensor([[[21, 2, 0, 5, 19], [10, 6, 10, 1, 5],

记录_broadcastcodeseed = 0n_approx = 5num_cells = 4dimension = 3X = np.random.rand(n_approx, dimension)print(X, X.shape)kernel_points = np.random.rand(num_cells, dimension)print(kernel_points, kernel_points.shape)>>>[[0.26214758 0.051704

Ubuntu18.04_anaconda3_pytorch_pycharm（community） 1. anaconda31.1 下载1.2 安装1.3 配置环境变量2. pytorch2.1 创建环境2.2官网下载3. Pycharm(community)3.1 下载3.2 安装3.3 创建桌面快捷方式4. 总结默认已完成NVIDIA驱动、CUDA、CuCNN的配置本文章为自己的操作记录，仅供参考1. anaconda31.1 下载登陆官网1.2 安装进入安装目录，启动安装程序

Ubuntu18.04_NVIDA_Cuda_CuDNNNVIDA显卡驱动安装NVIDIA CudaCuDNNNVIDA显卡驱动在应用程序中找到 Software & Updates（软件和更新）选择一个驱动，推荐最近的版本（末尾数字更大），点击Apply changes（应用更改）重启电脑安装NVIDIA Cuda依次点击对应模块：出现如下信息：（截至2021/3/10）按照上述顺序出现错误，经尝试，我成功的方案如下：wget https://developer.do

Ubuntu 18.04 新手教程_火狐浏览器flash插件安装1.Flash Player官网下载Flash Player官网地址选择：Linux（64-bit）选择：（.tar.gz）- NPAPI选择NPAPI的说明：2.解压与安装_命令行方式2.1 打开命令行： ctrl + alt + t2.2 切换到Downloads文件夹cd Downloads2.3 解压tar -zxvf flash_player_npapi_linux.x86_64.tar.gz

我通过计算机自带的“卸载程序”，卸载CATIA时，出现“找不到指定文件”的情况，又不想下载相关第三方软件，遂尝试本地删除。首先删除安装目录下的文件其次删除以下目录中的 Dassault SystemesC:\ProgramDataC:\Program FilesC:\Users\用户名\AppData\LocalC:\Users\用户名\AppData\Roaming最后删除 部分的 注册表（右击新建中的CATIA相关）进入注册表“开始”–>“运行”–>输入“reg

网格生成过程步骤生成3×3网格grid_size = 3grid = np.arange(grid_size)a, b = np.meshgrid(grid, grid) a = [[0 1 2] [0 1 2] [0 1 2]] b = [[0 0 0] [1 1 1] [2 2 2]]x_offset，y_offsetx_offset = torch.FloatTensor(a).view(-1, 1) # -1表示自适应y_off

pytorch中张量的阶数理解推荐打开2个页面，对比原四阶张量理解各阶的对应关系。创建一个四阶张量：import torchx = torch.linspace(0,71,72).view(2,3,3,4)print(x)输出：tensor([[[[ 0., 1., 2., 3.], [ 4., 5., 6., 7.], [ 8., 9., 10., 11.]], [[12., 13.,

4维张量的实例创建4维张量import torchx = torch.linspace(0,71,72).view(2,3,3,4)print(x)4维张量输出：tensor([[[[ 0., 1., 2., 3.], [ 4., 5., 6., 7.], [ 8., 9., 10., 11.]], [[12., 13., 14., 15.], [16., 17., 18.