在python中标记图像以进行分类 [英] Label images in python for classification

问题描述

我在两个文件夹中有约1400张图像，其中文件夹名称是图像的标签.我遍历每个文件夹，并将其与标签一起添加到数组中.后来我将其传递给我的神经网络模型.

I have ~1400 images in two folder where folder name is the label for image. I am iterating over each folder and appending it in an array along with label. Later i am passing it to my neural network model.

for folder in files:
   sdir = dir + "\\"
   sdir = sdir + folder
   print("inside loop " +sdir)
   count = 0
   for image in os.listdir(sdir):
     img = cv2.imread(image)
     count = count + 1
     if count <= 450:
       x_train.append(img)
       y_train.append(folder)
     else:
       x_test.append(img)
       y_test.append(folder)

model = tf.keras.models.Sequential()
model.add(tf.keras.layers.Flatten())
model.add(tf.keras.layers.Dense(128,activation=tf.nn.relu))
model.add(tf.keras.layers.Dense(128,activation=tf.nn.relu))
model.add(tf.keras.layers.Dense(10,activation=tf.nn.softmax))

Traceback (most recent call last):
  File "testing.py", line 43, in <module>
    model.fit(x_train,y_train, epochs=5)
  File "C:\Python\lib\site-packages\tensorflow\python\keras\engine\training.py", line 776, in fit
    shuffle=shuffle)
  File "C:\Python\lib\site-packages\tensorflow\python\keras\engine\training.py", line 2266, in _standardize_user_data
    'array or a list of arrays. You passed: x=' + str(x))
ValueError: Please provide as model inputs either a single array or a list of arrays.

推荐答案

我认为您的问题是由于 files 可能是字符串列表，因此 folder (您将其作为标签传递到网络中)是一个字符串，而网络将需要整数标签.

I imagine that your issue is due to the fact that files is probably a list of strings and therefore that folder(which you are passing as a label to your network) is a string, whereas the network will want integer labels.

因此，根据您拥有的标签数量(两个，如果我理解正确的话)，您可以定义一个词典

Therefore, depending on how many labels you have (two, if I understand correctly), you could define a dictionary

labels_to_ints = dict(zip(files,[0,1]))

将字符串映射到整数标签，然后使用

to map your strings to integer labels, and then instead of y_train.append(folder), use

y_train.append(labels_to_ints[folder])

(以及相应的 y_test ).

当然，还有其他方法可以执行此映射.

Of course, there are other ways to perform this mapping.

这篇关于在python中标记图像以进行分类的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！