kangzengxin的博客

import numpy as npimport matplotlib.pyplot as pltdef normalization(data): rang=np.max(data)-np.min(data) return (data-np.min(data))/rangdef save_1_feature_map(feature_map): # [1, H, W] -> [H, W] im = np.squeeze(feature_map.detach()

img为tensorimport torchimport numpy as npfrom PIL import Imageimport torchvision.transforms as transformsdef save_original_picture(img): # [3, H, W] img = transform_convert(img) # plt.figure() plt.axis('off') plt.imshow(img) p

查看pytorch版本import torchprint(torch.__version__) #注意是双下划线查看torchvision版本import torchvisiontorchvision.__version__gpu是否可用torch.cuda.is_available()是否检测到cudaprint(torch.version.cuda)

pybind11在Windows下的使用转自：https://www.e-learn.cn/content/qita/24527651.Pybind11算是目前最方便的Python调用C++的工具了, 介绍一下在vs2019上写Python的扩展的HelloWorldhttps://github.com/pybind/pybind11/releases/tag/v2.3.0这个库只要inc...