TypeError：预期的 str、bytes 或 os.PathLike 对象，而不是 Django 中的 JpegImageFile（或 PngImageFile）

Question

当我根据张量流h5 模型上传图片以检查图片时，我正在使用tensorflow.keras.models 的load_model 加载图像，但它不接受。对于 JPG，它显示为 TypeError: expected str, bytes or os.PathLike object, not JpegImageFile，对于 PNG，它显示为 TypeError: expected str, bytes or os.PathLike object, not PngImageFile。现在该怎么办？

我用原始 python 尝试了代码，但效果很好。

代码：

#views.py

import numpy as np
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing import image

pic = request.FILES['image']
img = Image.open(pic)
detection=load_model(os.path.join(settings.BASE_DIR,'static/auto_chloro_model.h5'))
test_img=image.load_img(img,target_size=(48,48))
test_img=image.img_to_array(test_img)
test_img=np.expand_dims(test_img,axis=0) 
result=detection.predict(test_img)
a=result.argmax()
print(a)

#models.py

class images(models.Model):
   img_main = models.ImageField(upload_to="images_api", default="") 

   def __str__(self):
        return self.product_name

#forms.py

class imageForm(forms.ModelForm):
    image = forms.ImageField()
  
    class Meta:
        model = images
        fields = ['image']

追溯： Internal Server Error: /upload Traceback (most recent call last): File "C:\Users\joyan\.conda\envs\tensorflow-django\lib\site-packages\django\core\handlers\exception.py", line 47, in inner response = get_response(request) File "C:\Users\joyan\.conda\envs\tensorflow-django\lib\site-packages\django\core\handlers\base.py", line 181, in _get_response response = wrapped_callback(request, *callback_args, **callback_kwargs) File "H:\Projects + Programming Projects\Auto Chloro\plant\detection\views.py", line 50, in uploadImage test_img=image.load_img(img,target_size=(48,48)) File "C:\Users\joyan\.conda\envs\tensorflow-django\lib\site-packages\keras_preprocessing\image\utils.py", line 113, in load_img with open(path, 'rb') as f: TypeError: expected str, bytes or os.PathLike object, not JpegImageFile

Answer 1

load_img 函数需要图像的路径，而不是 PIL 图像格式。

所以不是

img = Image.open(pic)
test_img=image.load_img(img,target_size=(48,48))

您可以保存图像，然后获取图像的 url/路径，然后使用以下命令提供image.load_img 的路径：

from tensorflow.keras.preprocessing import image
from django.core.files.storage import default_storage


# test_img = image.load_img('/path/to/image.jpg', target_size = (48, 48))
file_name = "pic.jpg"   # The file name with which it saves
file = default_storage.save(file_name, f)
file_url = default_storage.url(file)
test_img = image.load_img(file_url, target_size=(48, 48))

Answer 2

这解决了我的问题。

pic = request.FILES['image']
img_new = images(img_main= pic)
img_new.save()
detection=load_model(os.path.join(settings.BASE_DIR,'staticfiles/auto_chloro_model.h5'))
test_img=image.load_img(img_new.img_main.path,target_size=(48,48))
test_img=image.img_to_array(test_img)
test_img=np.expand_dims(test_img,axis=0) 
result=detection.predict(test_img)
a=result.argmax()