0. 图像批量处理一步到位：转格式(.jpg转.png)、尺寸(640*480)、位深度（RGB）等

import os
import glob
import os.path
from PIL import Image
from PIL import ImageFile        #批量修改尺寸2048*1024（此时也可能32位）
ImageFile.LOAD_TRUNCATED_IMAGES = Truepath = "G:\\Aerial\\"
files = os.listdir(path)          #会按顺序排列1,2,3,4，.png格式
i=0
for file in files:original = path + os.sep+ files[i]new = path + os.sep + str(i+1) + ".png"os.rename(original,new)i +=1
def Resize(file, outdir, width, height):imgFile = Image.open(file)try:newImage = imgFile.resize((width, height), Image.BILINEAR)newImage.save(os.path.join(outdir, os.path.basename(file)))except Exception as e:print(e)i = 0
a = 0
for file in glob.glob("G:\\Aerial\\*.png"): # 图片所在的目录# print(file)i += 1Resize(file, path, 640, 480)img = Image.open(file).convert('RGB')# 新图片存放的目录img.save(file)img1 = Image.open(file)# print(img1.getbands())  # ('P',) 这种是有彩色的，而L是没有彩色的# print(img1.size)if len(img1.getbands()) == 3:a +=1print(file)print('需要重新训练的图片数量：', i)
print('总图像是RGB的数量：', a)

检查下是否转化成功：

import os.path
from PIL import Image
from PIL import ImageFile        #批量修改尺寸2048*1024（此时也可能32位）
ImageFile.LOAD_TRUNCATED_IMAGES = True
import cv2path = "G:\\000\\image_ground150\\"files = os.listdir(path)
print(files)
i = 0
a = 0
c = 0
for pic in files:img = Image.open(os.path.join(path, pic))print(pic)print(img.getbands())  # ('P',) 这种是有彩色的，而L是没有彩色的print(img.size)i +=1# Img = np.array(img)# a = np.unique(Img)# print(a)    #看像素值if len(img.getbands()) == 3:a +=1size = img.sizew = size[0]  # 宽度h = size[1]  # 高度if w == 640:if h == 480:c += 1print('图像image的总数量: ', i)
print('总图像是RGB的数量：', a)
print('总图像是640*480的数量：', c)

1.批量排列1.2.3.4 统一成.png格式

import os
path = "F:/1207garbage classification/unknown"
files = os.listdir(path)          #会按顺序排列1,2,3,4，.png格式
i=0
for file in files:original = path + os.sep+ files[i]new = path + os.sep +"N12_"+ str(i+1) + ".png"os.rename(original,new)i+=1

有时候可能直接jpg转png就行，不用排序，转完格式和原图文件名保持一致：

# 以jpg转png为例，其他格式同理，
# 代码中路径更改为自己图像存放路径即可
from PIL import Image
import osimagesDirectory=r"D:\Images"  # jpg图片所在文件夹路径
distDirectory = os.path.dirname(imagesDirectory)# 保证jpg图像文件夹与png图像文件夹在同一目录下
distDirectory = os.path.join(distDirectory, "A")# 要存放png格式的文件夹路径
for imageName in os.listdir(imagesDirectory):imagePath = os.path.join(imagesDirectory, imageName)image = Image.open(imagePath)# 打开jpg图像distImagePath = os.path.join(distDirectory, imageName[:-4]+'.png')# 更改图像后缀为.png，并保证与原图像同名image.save(distImagePath)# 保存png图像

2.批量修改尺寸

import glob
import os.path
from PIL import Image
from PIL import ImageFile        #批量修改尺寸2048*1024（此时也可能32位）
ImageFile.LOAD_TRUNCATED_IMAGES = Truedef Resize(file, outdir, width, height):imgFile = Image.open(file)try:newImage = imgFile.resize((width, height), Image.BILINEAR)newImage.save(os.path.join(outdir, os.path.basename(file)))except Exception as e:print(e)for file in glob.glob("A/*.png"):  # 图片所在的目录Resize(file, "1/", 2048, 1024)  # 新图片存放的目录

3.批量统一到RGB三通道

from PIL import Image   #改成统一RGB通道
import ospath = "2"
save_path = "3"files = os.listdir(path)
print(files)for pic in files:img = Image.open(os.path.join(path, pic)).convert('RGB')print(pic)print(img.getbands())  # ('P',) 这种是有彩色的，而L是没有彩色的print(img.size)pic_new = os.path.join(save_path, pic)img.save(pic_new)

另一版本：

import glob
import os.path
import numpy as np
from PIL import Image
from PIL import ImageFile        #批量修改尺寸2048*1024（此时也可能32位）
ImageFile.LOAD_TRUNCATED_IMAGES = Truepath = "3"files = os.listdir(path)
print(files)for pic in files:img = Image.open(os.path.join(path, pic))print(pic)print(img.getbands())  # ('P',) 这种是有彩色的，而L是没有彩色的print(img.size)Img = np.array(img)a = np.unique(Img)print(a)    #看像素值

4.输出结果顺序、格式、位深通道、尺寸

5.labelme_json转mask

一步到位：

import cv2
import numpy as np
import json
import os.path
from PIL import Image
import globcategory_types = ["unlabeled","water"]h = 480
w = 640path = "G:\\V5_water\\json87"
filenames = os.listdir(path)
# for filename in filenames:
for filename in glob.glob("G:\\V5_water\\json87\\*.json"):mask = np.zeros([h, w, 1], np.uint8)# json_all = "G:\\0906\\json\\1.json"  # 原json路径json_all = os.path.join(path, filename)with open(json_all, 'r') as f: #json文件路径label = json.load(f)shapes = label["shapes"]for shape in shapes:category = shape["label"]points = shape["points"]#填充points_array = np.array(points, dtype=np.int32)mask = cv2.fillPoly(mask, [points_array], category_types.index(category))# if category == "road":    #高亮显示其中一个标签图像#     mask = cv2.fillPoly(mask, [points_array], 1)# # else:# if category == "sidewalk":#     mask = cv2.fillPoly(mask,[points_array], 2)# str = filename(len(ff):-5)name_json = os.path.split(json_all)[-1]    # '1.json'name = str(name_json)[0:-len('.json')] # '1'savename = "G:\\V5_water\\mask87\\" + name + ".png" # 需要在此路径下新建一个mask文件夹，结果图保存在此文件夹下面cv2.imwrite(savename, mask)img = Image.open(savename)Img = np.array(img)a = np.unique(Img)print(savename)print(mask.shape)print(img.getbands())print('像素值：',a)

不放心的话可以再试一次看看mask可视化效果：

import matplotlib.pyplot as plt
from matplotlib import gridspec
import numpy as np
import cv2
import osdef create_pascal_label_colormap():colormap = np.zeros((256, 3), dtype=int)ind = np.arange(256, dtype=int)for shift in reversed(range(8)):for channel in range(3):colormap[:, channel] |= ((ind >> channel) & 1) << shiftind >>= 3return colormapdef label_to_color_image(label):if label.ndim != 2:raise ValueError('Expect 2-D input label')colormap = create_pascal_label_colormap()if np.max(label) >= len(colormap):raise ValueError('label value too large.')return colormap[label]def vis_segmentation(image, seg_map):"""输入图片和分割 mask 的可视化."""plt.figure(figsize=(15, 5))grid_spec = gridspec.GridSpec(1, 4, width_ratios=[6, 6, 6, 1])plt.subplot(grid_spec[0])plt.imshow(image)plt.axis('off')plt.title('input image')plt.subplot(grid_spec[1])seg_image = label_to_color_image(seg_map).astype(np.uint8)plt.imshow(seg_image)plt.axis('off')plt.title('segmentation map')plt.subplot(grid_spec[2])plt.imshow(image)plt.imshow(seg_image, alpha=0.5)plt.axis('off')plt.title('segmentation overlay')unique_labels = np.unique(seg_map)ax = plt.subplot(grid_spec[3])plt.imshow(FULL_COLOR_MAP[unique_labels].astype(np.uint8), interpolation='nearest')ax.yaxis.tick_right()plt.yticks(range(len(unique_labels)), LABEL_NAMES[unique_labels])plt.xticks([], [])ax.tick_params(width=0.0)plt.grid('off')# plt.imsave('G:\\1\\' + image)# plt.show()LABEL_NAMES = np.asarray(['background', 'water'])  # 假设只有两类
FULL_LABEL_MAP = np.arange(len(LABEL_NAMES)).reshape(len(LABEL_NAMES), 1)
FULL_COLOR_MAP = label_to_color_image(FULL_LABEL_MAP)img_path = 'G:\\V5_water\\image87\\'   # img原图路径
png_path = 'G:\\V5_water\\mask87\\'   # mask输出全黑路径
i = 0
for img in os.listdir(img_path):print(img)imgfile = img_path + imgpngfile = png_path + imgnew_path = "G:\\V5_water\\plt\\" + img    #保存新的验证集val.txt测试结果可视化路径！！！img = cv2.imread(imgfile, 1)img = img[:, :, ::-1]seg_map = cv2.imread(pngfile, 0)vis_segmentation(img, seg_map)plt.savefig(new_path)plt.close()i +=1print('可视化总数量：',i)
print('Done.')

结果可视化如下：

如果原图是jpg，需要生成png格式进行训练的话：

import matplotlib.pyplot as plt
from matplotlib import gridspec
import numpy as np
import cv2
import os
from PIL import Imagedef create_pascal_label_colormap():colormap = np.zeros((256, 3), dtype=int)ind = np.arange(256, dtype=int)for shift in reversed(range(8)):for channel in range(3):colormap[:, channel] |= ((ind >> channel) & 1) << shiftind >>= 3return colormapdef label_to_color_image(label):if label.ndim != 2:raise ValueError('Expect 2-D input label')colormap = create_pascal_label_colormap()if np.max(label) >= len(colormap):raise ValueError('label value too large.')return colormap[label]def vis_segmentation(image, seg_map):"""输入图片和分割 mask 的可视化."""plt.figure(figsize=(15, 5))grid_spec = gridspec.GridSpec(1, 4, width_ratios=[6, 6, 6, 1])plt.subplot(grid_spec[0])plt.imshow(image)plt.axis('off')plt.title('input image')plt.subplot(grid_spec[1])seg_image = label_to_color_image(seg_map).astype(np.uint8)plt.imshow(seg_image)plt.axis('off')plt.title('segmentation map')plt.subplot(grid_spec[2])plt.imshow(image)plt.imshow(seg_image, alpha=0.5)plt.axis('off')plt.title('segmentation overlay')unique_labels = np.unique(seg_map)ax = plt.subplot(grid_spec[3])plt.imshow(FULL_COLOR_MAP[unique_labels].astype(np.uint8), interpolation='nearest')ax.yaxis.tick_right()plt.yticks(range(len(unique_labels)), LABEL_NAMES[unique_labels])plt.xticks([], [])ax.tick_params(width=0.0)plt.grid('off')# plt.imsave('G:\\1\\' + image)# plt.show()LABEL_NAMES = np.asarray(['background', 'water'])  # 假设只有两类
FULL_LABEL_MAP = np.arange(len(LABEL_NAMES)).reshape(len(LABEL_NAMES), 1)
FULL_COLOR_MAP = label_to_color_image(FULL_LABEL_MAP)# img_path = 'E:\\add_water1102\\FloW_IMG(water_segementation)\\image640x320\\'   # img原图路径
img_path = 'E:\\add_water1102\\FloW_IMG(water_segementation)\\all_images\\'
png_path = 'E:\\add_water1102\\FloW_IMG(water_segementation)\\all_masks\\'   # mask输出全黑路径
i = 0
new_input_path = 'E:\\add_water1102\\FloW_IMG(water_segementation)\\all_image_png\\'  # 如果是jpg的话会有生成png的路径
plt_path = "E:\\add_water1102\\FloW_IMG(water_segementation)\\plt\\"if not os.path.exists(new_input_path):os.makedirs(new_input_path)for img in os.listdir(img_path):print(img)imgfile = img_path + imga = Image.open(imgfile)a = a.resize((640, 480), Image.BILINEAR)a.save(imgfile)if img.endswith('.jpg'):img = img.replace(".jpg", ".png")if not os.path.exists(new_input_path):os.makedirs(new_input_path)else:passpngfile = png_path + imgb = Image.open(pngfile)b = b.resize((640, 480), Image.BILINEAR)b.save(pngfile)a = cv2.imread(imgfile)cv2.imwrite(new_input_path + img, a)new_path = plt_path + img  # 保存新的验证集val.txt测试结果可视化路径！！！img = cv2.imread(new_input_path + img, 1)img = img[:, :, ::-1]seg_map = cv2.imread(pngfile, 0)vis_segmentation(img, seg_map)plt.savefig(new_path)plt.close()i += 1else:# 原图pngpngfile = png_path + imgb = Image.open(pngfile)b = b.resize((640, 480), Image.BILINEAR)b.save(pngfile)new_path = plt_path + img.split(".")[0] + ".png"    #保存新的验证集val.txt测试结果可视化路径！！！img = cv2.imread(imgfile, 1)img = img[:, :, ::-1]seg_map = cv2.imread(pngfile, 0)vis_segmentation(img, seg_map)plt.savefig(new_path)plt.close()i +=1print('可视化总数量：',i)
print('Done.')

另一种bat大法：

1.bat文件

for %%m in (G:\haha\1\*.json) do (
python G:\labelme_json.py %%m
)

python文件：

import cv2
import numpy as np
import json
import sysdef main(argv):category_types = ["unlabeled","road line","sidewalk line"]h = 480w = 640mask = np.zeros([h, w, 1], np.uint8)filename = argv[1]with open(filename, "r") as f: #json文件路径label = json.load(f)shapes = label["shapes"]for shape in shapes:category = shape["label"]points = shape["points"]#填充points_array = np.array(points, dtype=np.int32)mask = cv2.fillPoly(mask, [points_array], category_types.index(category))# if category == "B":    #高亮显示其中一个标签图像#     mask = cv2.fillPoly(mask, [points_array], 255)# else:#     mask = cv2.fillPoly(mask,[points_array],category_types.index(category))ff = "G:/0811json/json/" #原json路径str = filename[len(ff):-5]savename = "G:/0811json/mask/" + str + ".png" # 需要在此路径下新建一个mask文件夹，结果图保存在此文件夹下面print(savename)cv2.imwrite(savename, mask)#cv2.waitKey(0)if __name__ == '__main__':main(sys.argv)

6.增强后！生成train.txt和val.txt路径

# 导入数据增强工具
import Augmentor# 确定原始图像存储路径以及掩码文件存储路径
p = Augmentor.Pipeline("image200") # 原图和mask的标签要一一对应！！扩充后的图都在原图里面新生成output的文件夹！
p.ground_truth("mask200") #标签图mask# 图像旋转： 按照概率0.8执行，最大左旋角度10，最大右旋角度10
p.rotate(probability=0.8, max_left_rotation=10, max_right_rotation=10)# 图像左右互换： 按照概率0.5执行
p.flip_left_right(probability=0.5)# 图像放大缩小： 按照概率0.8执行，面积为原始图0.85倍
p.zoom_random(probability=0.3, percentage_area=0.85)# 最终扩充的数据样本数（200到2000张，扩充了十倍）
p.sample(2000)

随机种子（比例9:1）生成训练和验证集666：（最新版）

from __future__ import absolute_import, print_function, division
import os
import randomimg_root = '.\\1\\'train_labeltxt = ".\\1\\train.txt"val_labeltxt = ".\\1\\val.txt"def get_img_mask_from_dir(data_dir, split_rate=0.1, random_seed=666):""":param data_dir: 图片数据存放路径:param split_rate: 拆分验证集的比例:param random_seed: 随机种子:return:"""train_paths = []val_paths = []for sub_dir in os.listdir(data_dir):if sub_dir == 'a30':  # 原图文件夹sub_path = os.path.join(data_dir, sub_dir)image_names = os.listdir(sub_path)nrof_img = len(image_names)indx = list(range(nrof_img))random.shuffle(indx, random.seed(random_seed))mask_dir = 'b30'   # mask文件夹mask_path = os.path.join(data_dir, mask_dir)mask_names = os.listdir(mask_path)# 原图和MASK名字对应索引# train_paths.extend([(os.path.join(sub_dir, image_names[i]).replace('\\', '/'), os.path.join(mask_dir, f'{os.path.splitext(image_names[i])[0]}.png').replace('\\', '/')) for i in indx[int(nrof_img*split_rate):]])# 原图和MASK名字按顺序索引train_paths.extend([(os.path.join(sub_dir, image_names[i]).replace('\\', '/'), os.path.join(mask_dir,f'{os.path.splitext(mask_names[i])[0]}.png').replace('\\', '/')) for i in indx[int(nrof_img*split_rate):]])val_paths.extend([(os.path.join(sub_dir, image_names[i]).replace('\\', '/'), os.path.join(mask_dir, f'{os.path.splitext(mask_names[i])[0]}.png').replace('\\', '/')) for i in indx[:int(nrof_img*split_rate)]])else:continuereturn train_paths, val_pathsif __name__ == '__main__':train_path, val_path = get_img_mask_from_dir(img_root, split_rate=0.1, random_seed=666)print('训练集数量train_path:',len(train_path))print('验证集数量val_path:',len(val_path))with open(train_labeltxt, 'a') as f:for line in train_path:f.write(line[0]+','+line[1]+'\n')with open(val_labeltxt, 'a') as f:for line in val_path:f.write(line[0]+','+line[1]+'\n')

随机生成训练和验证集：（旧版本）

from __future__ import absolute_import, print_function, division
import os
import randomimg_root = 'G:/water/'train_labeltxt = "G:/water/train.txt"val_labeltxt = "G:/water/val.txt"def get_img_mask_from_dir(data_dir, split_rate=0.1, random_seed=666):""":param data_dir: 图片数据存放路径:param split_rate: 拆分验证集的比例:param random_seed: 随机种子:return:"""train_paths = []val_paths = []for sub_dir in os.listdir(data_dir):if sub_dir == 'test0':sub_path = os.path.join(data_dir, sub_dir)image_names = os.listdir(sub_path)nrof_img = len(image_names)indx = list(range(nrof_img))random.shuffle(indx, random.seed(random_seed))mask_dir = 'mask0'train_paths.extend([(os.path.join(sub_dir, image_names[i]).replace('\\', '/'), os.path.join(mask_dir, f'{os.path.splitext(image_names[i])[0]}.png').replace('\\', '/')) for i in indx[int(nrof_img*split_rate):]])val_paths.extend([(os.path.join(sub_dir, image_names[i]).replace('\\', '/'), os.path.join(mask_dir, f'{os.path.splitext(image_names[i])[0]}.png').replace('\\', '/')) for i in indx[:int(nrof_img*split_rate)]])else:continuereturn train_paths, val_pathsif __name__ == '__main__':train_path, val_path = get_img_mask_from_dir(img_root, split_rate=0.1, random_seed=666)print('train_path:',len(train_path))print('val_path:',len(val_path))with open(train_labeltxt, 'a') as f:for line in train_path:f.write(line[0]+','+line[1]+'\n')with open(val_labeltxt, 'a') as f:for line in val_path:f.write(line[0]+','+line[1]+'\n')

不用新建train.txt 自动生成（旧版本）

（旧版本）

import ostrain_image_path = 'image/'
train_label_path = 'mask/'label_list = []
image_list = []train_image_dir = sorted(os.listdir(train_image_path))
train_label_dir = sorted(os.listdir(train_label_path))#print(train_label_dir)
file = './train.txt'
with open(file, 'w') as f:for n1, n2 in zip(train_image_dir, train_label_dir):train_single_image = os.path.join(train_image_path, n1)train_single_label = os.path.join(train_label_path, n2)# print(train_single_label)f.write(train_single_image + ',' + train_single_label + '\n')f.close()

生成train.txt（旧版本）

7.用linux命令将.txt按照行打散

shuf 6.txt -o output.txt

打散前：

打散后：

可视化mask:

import matplotlib.pyplot as plt
from matplotlib import gridspec
import numpy as np
import cv2def create_pascal_label_colormap():colormap = np.zeros((256, 3), dtype=int)ind = np.arange(256, dtype=int)for shift in reversed(range(8)):for channel in range(3):colormap[:, channel] |= ((ind >> channel) & 1) << shiftind >>= 3return colormapdef label_to_color_image(label):if label.ndim != 2:raise ValueError('Expect 2-D input label')colormap = create_pascal_label_colormap()if np.max(label) >= len(colormap):raise ValueError('label value too large.')return colormap[label]def vis_segmentation(image, seg_map):"""输入图片和分割 mask 的可视化."""plt.figure(figsize=(15, 5))grid_spec = gridspec.GridSpec(1, 4, width_ratios=[6, 6, 6, 1])plt.subplot(grid_spec[0])plt.imshow(image)plt.axis('off')plt.title('input image')plt.subplot(grid_spec[1])seg_image = label_to_color_image(seg_map).astype(np.uint8)plt.imshow(seg_image)plt.axis('off')plt.title('segmentation map')plt.subplot(grid_spec[2])plt.imshow(image)plt.imshow(seg_image, alpha=0.7)plt.axis('off')plt.title('segmentation overlay')unique_labels = np.unique(seg_map)ax = plt.subplot(grid_spec[3])plt.imshow(FULL_COLOR_MAP[unique_labels].astype(np.uint8), interpolation='nearest')ax.yaxis.tick_right()plt.yticks(range(len(unique_labels)), LABEL_NAMES[unique_labels])plt.xticks([], [])ax.tick_params(width=0.0)plt.grid('off')plt.show()LABEL_NAMES = np.asarray(['background', 'water']) # 假设只有两类
FULL_LABEL_MAP = np.arange(len(LABEL_NAMES)).reshape(len(LABEL_NAMES), 1)
FULL_COLOR_MAP = label_to_color_image(FULL_LABEL_MAP)imgfile = '1.png'
pngfile = '2.png'
img = cv2.imread(imgfile, 1)
img = img[:,:,::-1]
seg_map = cv2.imread(pngfile, 0)
vis_segmentation(img, seg_map)print('Done.')

20210810 所有图像数据准备一条龙（labelme_json转mask、数据增强Augmentor、随机种子设比例生成train.val、转格式(.jpg转.png)、尺寸、位深度变换相关推荐

Radiology：影像组学：图像不仅仅是图片，它们还是数据
在过去十年中,随着模式识别方法的增加和数据集大小的增加,医学图像分析领域的研究越来越多.这些研究促进了对图像定量特征高通量提取过程的发展,从而将图像转换为可挖掘数据特征,并随后对这些数据进行分析,以提 ...
（1）【数据隐藏】一起入门隐写吧，宝？word、图像、移动设备、文件压缩数据隐藏
目录第2章数据隐藏简单练习4则 2.1 在word中隐藏数据如何在word中隐藏数据?(包含不易被发现的傻瓜方法) 如何设置显示隐藏文字?(微软自带) 如何检测隐藏在文档里的信息?(避 ...
JPG图像exif和XPM信息中GPS数据姿态数据航向角数据的提取
JPG图像的编码相关内容太多不在多说了大家随手能查很多资料. 今天重点说说图像数据中的GPS信息以及飞机.相机姿态角度数据提取. JPG作为复杂的图像数据很多人都知道存在一个叫做EXIF的数据规范,在 ...
图像位深度 8位 16位 24位 32位区别对比 RGB 真彩色基本概念：（大小，深度，通道）位深度数据类型转换原理 Mat数据读取（opencv里的imread）
位深度位深度是指在记录数字图像的颜色时,计算机实际上是用每个像素需要的二进制数值位数来表示的.计算机之所以能够显示颜色,是采用了一种称作"位"( bit ) 的记数单位来记录所表 ...
使用 LJ1-01 夜间灯光图像检测县域经济发展：与 NPP-VIIRS 数据的比较
抽象的夜间灯光(NTL)遥感数据已被广泛用于推导国家和区域尺度的社会经济指标,以研究区域经济发展.然而,以往的研究大多只选择单一的衡量指标(如GDP),采用简单的回归方法,基于DMSP-OLS或NP ...
图像搜索引擎1|使用Python颜色直方图在数据集中搜索视觉上相似的图像
图像搜索引擎1|使用Python颜色直方图在数据集中搜索视觉上相似的图像这篇博客将介绍如何从头到尾创建图像搜索引擎.第一步是选择一个图像描述符--使用3D RGB直方图来表征图像的颜色.然后通过提取 ...
耕耘数据，融合发展——2018年度数据科学研究院RONG教授座谈会成功举办
2018年05月07日,以"耕耘数据.融合发展"为主题的2018年度数据科学研究院(以下简称"数据院")RONG教授座谈会于双清大厦拉开帷幕.数据院院长俞士纶. ...
《企业大数据系统构建实战：技术、架构、实施与应用》——第3章企业大数据解决方案 3.1 企业大数据解决方案实现方式...
本节书摘来自华章计算机<企业大数据系统构建实战:技术.架构.实施与应用>一书中的第3章,第3.1节,作者吕兆星郑传峰宋天龙杨晓鹏,更多章节内容可以访问云栖社区"华章计算机 ...
四说大数据时代“神话”：从大数据到深数据\n
作为国内最大的电商平台之一,苏宁每天要处理数量巨大的数据.为了更快速高效地处理这些数据,苏宁调度平台采取了哪些措施呢? 本文是苏宁大数据离线任务开发调度平台实践系列文章之上篇,详解苏宁的任务调度模块. ...

20210810 所有图像数据准备一条龙（labelme_json转mask、数据增强Augmentor、随机种子设比例生成train.val、转格式(.jpg转.png)、尺寸、位深度变换