m0_37908025的博客

torch.onnx.export（）函数中export_params=True，不能是False。

#define _CRT_SECURE_NO_WARNINGS#include <stdio.h> int main() { char *filename = "E://darknet-master-cpu/build/ceshi.txt"; //文件名 FILE *fp; char name[256]; //每行最大读取的字符数 fp = fopen(filename, "r");

import osfile_path = "Annotations"# os.listdir(file)会历遍文件夹内的文件并返回一个列表path_list = os.listdir(file_path)# print(path_list)# 定义一个空列表,我不需要path_list中的后缀名path_name=[]# 利用循环历遍path_list列表并且利用split去掉后缀名for i in path_list: path_name.append(i.split(".")[

import cv2import osrootdir = r'./before/' # 指明被遍历的文件夹for parent, dirnames, filenames in os.walk(rootdir): for filename in filenames: currentPath = os.path.join(parent, filename) img = cv2.imread(currentPath) cropped = img[1920:3360, 0:1440] #

clc;clear;namelist = dir('F:\mat\*.png');len = length(namelist); for i = 1:len png_file = fullfile( 'F:\mat\', namelist(i).name ); I = imread( png_file); % path = 'F:\mat\' name = '001' ext = '.png' [ path , name , ext ] = filep

import osfrom PIL import Imageimport cv2 as cvimport numpy as nplabel_dir = './label/'label_process_dir = './label_process/'if not os.path.exists(label_process_dir): os.makedirs(label_process_dir)for file in os.listdir(label_dir): label =

import numpy as npimport osimport cv2 as cvlabel_dir = './png/'label_process_dir = './label_process/'for file in os.listdir(label_dir): label = label_dir + file if os.path.isfile(label): print(file) # 获得一张图片的所有种类的灰度值 l

from PIL import Imageimport numpy as npimport osimport os.pathimport cv2MAX_VALUE = 100'''### 图片左右对称 ===rootdir = r'F:\data\png_3' # 指明被遍历的文件夹for parent, dirnames, filenames in os.walk(rootdir): for filename in filenames: print('parent i

def mask2onehot(mask, num_classes): # mask为灰度图，(H,W)→(K,H,W) _mask = [mask == i for i in range(num_classes)] return np.array(_mask).astype(np.uint8)

from visdom import Visdomimport scipy.io as ioimport numpy as np#python -m visdom.server #于cmd输入vis = Visdom()pred = io.loadmat(‘E:/abc/123.mat’)pred = pred[‘pred’].astype(np.uint8) #'pred’为自定义的.mat内部变量pred = 100 * pred #100为了拉大灰度值差异，可自定义vis.im

import torchimport torch.nn as nnimport torchvision.datasets as normal_datasetsimport torchvision.transforms as transformsfrom torch.autograd import Variablenum_epochs = 5learning_rate = 0.001def get_variable(x): x = Variable(x) return x.cuda()