首先先把数据集的图片路径保存在一个txt文件夹里面
import os def generate(dir, label): listText = open( 'list.txt', 'a') for file in dir: fileType = os.path.split(file) if fileType[ 1] == '.txt': continue name = file + ' ' + str(int(label)) + '\n' listText.write(name) listText.close() outer_path = 'E:/lly/data/' # 这里是你的图片的目录 if __name__ == '__main__': i = 1 num = 0 personlist = os.listdir(outer_path) # 列举文件夹 personlist.sort() for person in personlist: personPath = outer_path+person + "/" fingerlist = os.listdir(personPath) fingerlist.sort() for finger in fingerlist: finallPATH=os.path.join(outer_path, person,finger) finallPATH=finallPATH.replace( '\\', '/') listText = open( 'image_list.txt', 'a') fileType = os.path.split(finallPATH) name = finallPATH+ '\n' listText.write(name) listText.close() i += 1
计算自己数据集的均值和方差:
# -*- coding: utf-8 -*-** import numpy as np import cv2 import random import os # calculate means and std 注意换行\n符号** # train.txt中每一行是图像的位置信息** path = 'C:/Users/lenovo/PycharmProjects/my/image_list.txt' means = [ 0, 0, 0] stdevs = [ 0, 0, 0] index = 1 num_imgs = 0 with open(path, 'r') as f: lines = f.readlines() # random.shuffle(lines) for line in lines: print(line) print( '{}/{}'.format(index, len(lines))) index += 1 a = os.path.join(line) # print(a[:-1]) num_imgs += 1 img = cv2.imread(a[: -1]) print(img, 22) img = np.asarray(img) img = img.astype(np.float32) / 255. for i in range( 3): means[i] += img[:, :, i].mean() stdevs[i] += img[:, :, i].std() print(num_imgs) means.reverse() stdevs.reverse() means = np.asarray(means) / num_imgs stdevs = np.asarray(stdevs) / num_imgs print( "normMean = {}".format(means)) print( "normStd = {}".format(stdevs)) print( 'transforms.Normalize(normMean = {}, normStd = {})'.format(means, stdevs))
参考:https://blog.csdn.net/weixin_38533896/article/details/85951903