我的数据集中有一个包含分类字符串的特殊功能。这些值属于['a', 'ae', 'e', 'i', 'u']
但是，我想将这些字符映射为数字。请注意，我使用的是tensorflow数据集。
这是我的示例代码：
data_dir = "C:/Users/user/Documents/vowels/"
# I have data collected from 13 different subjects. Each time the data is recorded is considered one trial. In total we have 6 trials per subject.
# In this case, I used the first 5 trials for training and the 6th for testing/validation.
subjects_nums = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13]
trial_nums_train = [1, 2, 3, 4, 5]
trial_nums_test = [6]
paths_train = [data_dir + 'Col3/*/*_{}_trail_{}.png'.format(i, j) for i in subjects_nums for j in trial_nums_train]
paths_test = [data_dir + 'Col3/*/*_{}_trail_{}.png'.format(i, j) for i in subjects_nums for j in trial_nums_test]
list_ds_train = tf.data.Dataset.list_files(paths_train)
list_ds_test = tf.data.Dataset.list_files(paths_test)
# Here in this case, I did the conversion manually, on purpose for now. However, what if I don't know all the categories, or if I have 10s of them. I would like to convert the strings into numbers automatically.
def get_label(file_path):
# convert the path to a list of path components
parts = tf.strings.split(file_path, os.path.sep)
# The second to last is the class-directory
char = tf.strings.split(parts[-2], "_")[1]
tensor = char
if tensor == 'a':
return 0
elif tensor == 'ae':
return 1
elif tensor == 'e':
return 2
elif tensor == 'i':
return 3
else:
return 4
def decode_img(img):
# convert the compressed string to a 3D uint8 tensor
img = tf.image.decode_jpeg(img, channels=3)
# Use `convert_image_dtype` to convert to floats in the [0,1] range.
img = tf.image.convert_image_dtype(img, tf.float32)
# resize the image to the desired size.
return img
def process_path(file_path):
label = get_label(file_path)
# load the raw data from the file as a string
img = tf.io.read_file(file_path)
img = decode_img(img)
return img, label
# Use Dataset.map to create a dataset of image, label pairs:
# Set `num_parallel_calls` so multiple images are loaded/processed in parallel.
AUTOTUNE = tf.data.experimental.AUTOTUNE
labeled_ds_train = list_ds_train.map(process_path, num_parallel_calls=AUTOTUNE)
labeled_ds_test = list_ds_test.map(process_path, num_parallel_calls=AUTOTUNE)
labeled_ds_train = labeled_ds_train.cache().shuffle(buffer_size=1000).batch(32).prefetch(AUTOTUNE)
labeled_ds_test = labeled_ds_test.cache().batch(32).prefetch(AUTOTUNE)
然后检查数据集包含的内容：
for image, label in labeled_ds_train.take(1):
print("Image shape: ", image.numpy().shape)
print("Label: ", label.numpy())
我知道了：
Image shape: (32, 130, 267, 3)
Label: [b'ae' b'u' b'a' b'e' b'i' b'ae' b'i' b'e' b'e' b'a' b'i' b'a' b'i' b'a' b'i' b'u' b'u' b'ae' b'e' b'a' b'e' b'ae' b'a' b'i' b'i' b'e' b'ae' b'i' b'i' b'e' b'e' b'i']
我想要一种简单的方法，可以随时或自动将字符串转换为数字。
那怎么可能？
同样，首先，我有一个名为vowels的根文件夹，然后有一个名为Col3和Col4的子文件夹。然后，它们每个都包含子文件夹vowel_a，vowel_ae，vowel_e，vowel_i和vowel_u。然后将图像存储在后面的子文件夹中。图片名称如下：subject _ {} trial {}。png;其中第一个持有人反映主题编号，第二个持有人反映主题审判。
非常感谢您的帮助