《深度学习——实战caffe》——caffe可视化方法python版

（python3.6.4+caffe+pycharm）

一、数据可视化

1、mnist数据可视化

      1）、训练样本可视化

      首先要打开pycharm

      新建python文件1.py，代码输入：

# -*- coding: utf-8 -*
import numpy as np
import struct

from PIL import Image
import os

data_file = '/caffe/data/mnist/train-images-idx3-ubyte'  # 需要修改的路径，train-images-idx3-ubyte文件所在的位置
# It's 47040016B, but we should set to 47040000B
data_file_size = 47040016
data_file_size = str(data_file_size - 16) + 'B'

data_buf = open(data_file, 'rb').read()

magic, numImages, numRows, numColumns = struct.unpack_from(
    '>IIII', data_buf, 0)
datas = struct.unpack_from(
    '>' + data_file_size, data_buf, struct.calcsize('>IIII'))
datas = np.array(datas).astype(np.uint8).reshape(
    numImages, 1, numRows, numColumns)

label_file = '/caffe/data/mnist/train-labels-idx1-ubyte'  # 需要修改的路径 ，train-images-idx3-ubyte文件所在位置，最好采用绝对路径
# It's 60008B, but we should set to 60000B
label_file_size = 60008
label_file_size = str(label_file_size - 8) + 'B'

label_buf = open(label_file, 'rb').read()

magic, numLabels = struct.unpack_from('>II', label_buf, 0)
labels = struct.unpack_from(
    '>' + label_file_size, label_buf, struct.calcsize('>II'))
labels = np.array(labels).astype(np.int64)

datas_root = '/home/jinhanjun/caffe/examples/mnist/mnist_train'  # 需要修改的路径，你最终可视化后的图片保存在哪里
if not os.path.exists(datas_root):
    os.mkdir(datas_root)

for i in range(10):
    file_name = datas_root + os.sep + str(i)
    if not os.path.exists(file_name):
        os.mkdir(file_name)

for ii in range(numLabels):
    img = Image.fromarray(datas[ii, 0, 0:28, 0:28])
    label = labels[ii]
    file_name = datas_root + os.sep + str(label) + os.sep + 'mnist_train_' + str(ii) + '.png'
    img.save(file_name)

  运行上面程序，可得到训练用的50000个样本集图片。打开/home/jinhanjun/caffe/examples/mnist/mnist_train文件即可查看。

2)、测试样本可视化

        在pycharm新建python程序2.py

import numpy as np
import struct

from PIL import Image
import os

data_file = '/caffe/data/mnist/t10k-images-idx3-ubyte'  # 需要修改的路径，t10k-images-idx3-ubyte文件所在的位置

# It's 7840016B, but we should set to 7840000B
data_file_size = 7840016
data_file_size = str(data_file_size - 16) + 'B'

data_buf = open(data_file, 'rb').read()

magic, numImages, numRows, numColumns = struct.unpack_from(
    '>IIII', data_buf, 0)
datas = struct.unpack_from(
    '>' + data_file_size, data_buf, struct.calcsize('>IIII'))
datas = np.array(datas).astype(np.uint8).reshape(
    numImages, 1, numRows, numColumns)

label_file = '/caffe/data/mnist/t10k-labels-idx1-ubyte'  # 需要修改的路径，标签t10k-labels-idx1-ubyte文件所在位置
# It's 10008B, but we should set to 10000B
label_file_size = 10008
label_file_size = str(label_file_size - 8) + 'B'
label_buf = open(label_file, 'rb').read()
magic, numLabels = struct.unpack_from('>II', label_buf, 0)
labels = struct.unpack_from(
    '>' + label_file_size, label_buf, struct.calcsize('>II'))
labels = np.array(labels).astype(np.int64)
datas_root = '/home/jinhanjun/caffe/examples/mnist/mnist_test'  # 需要修改的路径（可视化后保存的位置）
if not os.path.exists(datas_root):
    os.mkdir(datas_root)

for i in range(10):
    file_name = datas_root + os.sep + str(i)
    if not os.path.exists(file_name):
        os.mkdir(file_name)

for ii in range(numLabels):
    img = Image.fromarray(datas[ii, 0, 0:28, 0:28])
    label = labels[ii]
    file_name = datas_root + os.sep + str(label) + os.sep + 'mnist_test_' + str(ii) + '.png'
    img.save(file_name)

  运行上面程序，在相应的文件/home/jinhanjun/caffe/examples/mnist/mnist_test中查看

 2、cifar10数据可视化    

首先下载python版cifar10数据。

先给个cifar数据下载链接:http://www.cs.toronto.edu/~kriz/cifar.html      

链接上提到三个数据版本,分别是python,matlab,binary版本,分别适合python,matlab,C程序       

下载cifar-10-python.tar.gz文件，下载下来复制到caffe/data/cifar10文件夹中，解压待用。

  然后就是pycharm写代码来运行程序了。代码如下：      

import pickle as p
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.image as plimg
from PIL import Image

def load_CIFAR_batch(filename):
    """ load single batch of cifar """
    with open(filename, 'rb')as f:
        datadict = p.load(f,encoding='iso-8859-1')
        X = datadict['data']
        Y = datadict['labels']
        X = X.reshape(10000, 3, 32, 32)
        Y = np.array(Y)
        return X, Y

def load_CIFAR_Labels(filename):
    with open(filename, 'rb') as f:
        lines = [x for x in f.readlines()]
        print(lines)

if __name__ == "__main__":
    load_CIFAR_Labels("/home/jinhanjun/caffe/data/cifar-10-batches-py/batches.meta") #batches.meta路径,刚下载下来的cifar10数据文件夹中包含
    imgX, imgY = load_CIFAR_batch("/home/jinhanjun/caffe/data/cifar-10-batches-py/data_batch_1")  #data_batch_1路径，刚下载下来的cifar10数据文件中包含
    print(imgX.shape)
    print("正在保存图片:")
    for i in range(imgX.shape[0]):
        imgs = imgX[i - 1]
        if i < 100:#只循环100张图片,这句注释掉可以便利出所有的图片,图片较多,可能要一定的时间
            img0 = imgs[0]
            img1 = imgs[1]
            img2 = imgs[2]
            i0 = Image.fromarray(img0)
            i1 = Image.fromarray(img1)
            i2 = Image.fromarray(img2)
            img = Image.merge("RGB",(i0,i1,i2))
            name = "img" + str(i)
            img.save("/home/jinhanjun/caffe/examples/images/cifar10/images/"+name,"png")#文件夹下是RGB融合后的图，保存的路径，需要特别注意的一点，此路径如果是要保存在你原本没有建立的文件夹下的情况下，需要自己手动建立，不像前面mnist程序会自己建立，而这个程序运行是不会自动建立的，如果你没有建立，程序会报错，显示路径问题。
            for j in range(imgs.shape[0]):
                img = imgs[j - 1]
                name = "img" + str(i) + str(j) + ".png"
                print("正在保存图片" + name)
                plimg.imsave("/home/jinhanjun/caffe/examples/images/cifar10/image/" + name, img)#文件夹下是RGB分离的图像，保存的图像路径，同上面所说的，注意路径的建立。

    print("保存完毕.")

我们可以在/home/jinhanjun/caffe/examples/images/cifar10/images/文件夹下和/home/jinhanjun/caffe/examples/images/cifar10/image/文件夹下查看保存的图片，后者图片数量是前者的三倍