英文:
Converting multiple images into tensor at once
问题
以下是翻译好的部分：
我有大约10张图片。这些图片大小各不相同。我正在使用OpenCV读取它们并将它们全部转换为（54, 54, 3）的格式。我知道可以将它们每个都转换为张量。有没有一种方法可以将它们一起转换为张量？最终的张量将具有形状（10, 54, 54, 3）。
我正在尝试使用以下代码进行单张图片的转换。
import cv2, glob
import tensorflow as tf
lst = glob.glob("images/*jpg")
im1 = cv2.imread(lst[1])
im1_tensor = tf.convert_to_tensor(im1, dtype=tf.int64)
英文:
I have ~10 images. They images are different in size. I am reading them in OpenCV and converting all of them to (54, 54, 3). I know that I can convert each one of them to a tensor. Is there any way of converting them to tensors altogether? Final tensor will have a shape of (10, 54, 54, 3)
I am trying to use the following codes for single image conversion.
import cv2, glob
import tensorflow as tf
lst = glob.glob("images/*jpg")
im1 = cv2.imread(lst[1])
im1_tensor = tf.convert_to_tensor(im1, dtype=tf.int64)
答案1
得分: 1
尝试以下代码以进行批处理处理。我已经使用了tf.data API。
def parse_image(filepath):
# 在这个函数中编写您自己的解析逻辑
image = tf.io.read_file(filepath)
image = tf.io.decode_jpeg(image)
image = tf.io.encode_jpeg(image)
return image
batch_size = 10
pred_files_ds = tf.data.Dataset.list_files(str(IMAGE_PATH/'*'))
pred_imgs_ds = pred_files_ds.map(lambda x: parse_image(x))
for img_batch in pred_imgs_ds.batch(batch_size):
print(img_batch.shape)
英文:
Try the following code to do batch processing. I have used tf.data API.
def parse_image(filepath):
# Write your own parsing logic in this function
image = tf.io.read_file(filepath)
image = tf.io.decode_jpeg(image)
image = tf.io.encode_jpeg(image)
return image
batch_size = 10
pred_files_ds = tf.data.Dataset.list_files(str(IMAGE_PATH/"*"))
pred_imgs_ds = pred_files_ds.map(lambda x: parse_image(x))
for img_batch in pred_imgs_ds.batch(batch_size):
print(img_batch.shape)
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论