AttributeError: 'NoneType' object has no attribute 'shape'

My data is organised as:
train:- containing folder for each class
valid :- containing folder for each class
test:- contains all test images


I have the same issue.

Without seeing the entire error message, my first guess would be to delete the tmp folder and then try again.

I have the same issue with the lesson1 notebook. It works fine using the dogscats dataset, but when I use another dataset with dog breeds, it fails with the same error as above. Deleting the tmp directory doesn’t help. This is the entire stack trace:

AttributeError                            Traceback (most recent call last)
<ipython-input-17-42d5f498bf97> in <module>()
      1 arch=resnet34
      2 data = ImageClassifierData.from_paths(PATH, tfms=tfms_from_model(arch, sz))
----> 3 learn = ConvLearner.pretrained(arch, data, precompute=True)
      4, 3)

~\fastai\fastai\ in pretrained(cls, f, data, ps, xtra_fc, xtra_cut, **kwargs)
     96     def pretrained(cls, f, data, ps=None, xtra_fc=None, xtra_cut=0, **kwargs):
     97         models = ConvnetBuilder(f, data.c, data.is_multi, data.is_reg, ps=ps, xtra_fc=xtra_fc, xtra_cut=xtra_cut)
---> 98         return cls(data, models, **kwargs)
    100     @property

~\fastai\fastai\ in __init__(self, data, models, precompute, **kwargs)
     89         elif self.metrics is None:
     90             self.metrics = [accuracy_multi] if else [accuracy]
---> 91         if precompute: self.save_fc1()
     92         self.freeze()
     93         self.precompute = precompute

~\fastai\fastai\ in save_fc1(self)
    135         m=self.models.top_model
    136         if len(self.activations[0])!=len(
--> 137             predict_to_bcolz(m,, act)
    138         if len(self.activations[1])!=len(
    139             predict_to_bcolz(m,, val_act)

~\fastai\fastai\ in predict_to_bcolz(m, gen, arr, workers)
     11     lock=threading.Lock()
     12     m.eval()
---> 13     for x,*_ in tqdm(gen):
     14         y = to_np(m(VV(x)).data)
     15         with lock:

c:\users\hille\anaconda3\envs\dlwin36\lib\site-packages\tqdm\ in __iter__(self)
    951 """, fp_write=getattr(self.fp, 'write', sys.stderr.write))
--> 953             for obj in iterable:
    954                 yield obj
    955                 # Update and possibly print the progressbar.

~\fastai\fastai\ in __next__(self)
    242         if self.i>=len(self.dl): raise StopIteration
    243         self.i+=1
--> 244         return next(
    246     @property

~\fastai\fastai\ in __iter__(self)
     73     def __iter__(self):
     74         with ThreadPoolExecutor(max_workers=self.num_workers) as e:
---> 75             for batch in, iter(self.batch_sampler)):
     76                 yield get_tensor(batch, self.pin_memory)

c:\users\hille\anaconda3\envs\dlwin36\lib\concurrent\futures\ in result_iterator()
    554                 for future in fs:
    555                     if timeout is None:
--> 556                         yield future.result()
    557                     else:
    558                         yield future.result(end_time - time.time())

c:\users\hille\anaconda3\envs\dlwin36\lib\concurrent\futures\ in result(self, timeout)
    396                 raise CancelledError()
    397             elif self._state == FINISHED:
--> 398                 return self.__get_result()
    400             self._condition.wait(timeout)

c:\users\hille\anaconda3\envs\dlwin36\lib\concurrent\futures\ in __get_result(self)
    355     def __get_result(self):
    356         if self._exception:
--> 357             raise self._exception
    358         else:
    359             return self._result

c:\users\hille\anaconda3\envs\dlwin36\lib\concurrent\futures\ in run(self)
     54         try:
---> 55             result = self.fn(*self.args, **self.kwargs)
     56         except BaseException as e:
     57             self.future.set_exception(e)

~\fastai\fastai\ in get_batch(self, indices)
     67     def get_batch(self, indices):
---> 68         res = self.collate_fn([self.dataset[i] for i in indices], self.pad_idx)
     69         if not self.transpose: return res
     70         res[0] = res[0].T

~\fastai\fastai\ in <listcomp>(.0)
     67     def get_batch(self, indices):
---> 68         res = self.collate_fn([self.dataset[i] for i in indices], self.pad_idx)
     69         if not self.transpose: return res
     70         res[0] = res[0].T

~\fastai\fastai\ in __getitem__(self, idx)
     96     def __getitem__(self, idx):
     97         x,y = self.get_x(idx),self.get_y(idx)
---> 98         return self.get(self.transform, x, y)
    100     def __len__(self): return self.n

~\fastai\fastai\ in get(self, tfm, x, y)
    102     def get(self, tfm, x, y):
--> 103         return (x,y) if tfm is None else tfm(x,y)
    105     @abstractmethod

~\fastai\fastai\ in __call__(self, im, y)
    463         if crop_type == CropType.NO: crop_tfm = NoCropXY(sz, tfm_y)
    464         self.tfms = tfms + [crop_tfm, normalizer, channel_dim]
--> 465     def __call__(self, im, y=None): return compose(im, y, self.tfms)

~\fastai\fastai\ in compose(im, y, fns)
    444 def compose(im, y, fns):
    445     for fn in fns:
--> 446         im, y =fn(im, y)
    447     return im if y is None else (im, y)

~\fastai\fastai\ in __call__(self, x, y)
    229     def __call__(self, x, y):
    230         self.set_state()
--> 231         x,y = ((self.transform(x),y) if self.tfm_y==TfmType.NO
    232                 else self.transform(x,y) if self.tfm_y==TfmType.PIXEL
    233                 else self.transform_coord(x,y))

~\fastai\fastai\ in transform(self, x, y)
    238     def transform(self, x, y=None):
--> 239         x = self.do_transform(x)
    240         return (x, self.do_transform(y)) if y is not None else x

~\fastai\fastai\ in do_transform(self, x)
    324     def do_transform(self, x):
--> 325         return scale_min(x,

~\fastai\fastai\ in scale_min(im, targ)
     10         targ (int): target size
     11     """
---> 12     r,c,*_ = im.shape
     13     ratio = targ/min(r,c)
     14     sz = (scale_to(c, ratio, targ), scale_to(r, ratio, targ))

AttributeError: 'NoneType' object has no attribute 'shape'

After debugging, I found that the issue is caused by having files other than jpg files in one of the image directories. There are several ways to fix this:

  1. Remove the unwanted files from the directories manually.
  2. Skip non-jpg files in the read_dirs function in
  3. Skip file if open_img returns None in FilesDataSet.get_x

I used option 2) and that solved my issue.


Hi Robert,

I’m having trouble implementing a fix for any of the 3 options. I wrote a function that, as far as I can tell, successfully converts all of my images to jpg files. Can you share what you did for your fix aimed at 2)?

Edit: Nevermind, spoke too soon!

This the function I changed for option 2:

def read_dirs(path, folder):
    labels, filenames, all_labels = [], [], []
    full_path = os.path.join(path, folder)
    for label in sorted(os.listdir(full_path)):
        for fname in os.listdir(os.path.join(full_path, label)):
            if fname.endswith('.jpg'):
                filenames.append(os.path.join(folder, label, fname))
    return filenames, labels, all_labels

The additional if statement simply skips files that don’t end with “.jpg”. There are probably better ways to do this.

1 Like

Hi I am also getting this error. I merely changed the path from data/dogscats/ to the one I created for my own data (data/dolphinsharks) and within it a validation and training set. Within that I split into each category of dolphins and sharks.

When I ran the below code, I got the attribute error: ‘NoneType’ object has no attribute ‘shape’. Any idea what I did wrong?

Edit: i also ensured the images uploaded were jpg.

data = ImageClassifierData.from_paths(PATH, tfms=tfms_from_model(arch, sz))
learn = ConvLearner.pretrained(arch, data, precompute=True), 3)

1 Like

Here’s my story with this error:
my data was polluted with undesirable files: hidden files (name starting with dot) came from my mac:
.DS_Store and that sort of business.
Also a mysterious photo file starting with dot underscore:
._bird_165.jpg instead of bird_165.jpg
I have no idea where that one came from or why it caused my error. Just look out for these filenames. A normal ls command will not show them unless you use the argument -a.

Do a la -ain the directory and check for unwanted files

I found the source of my unwanted ._ files:
macos’ tar puts them in, unless you follow instructions here:

mate could you clarify what you mentioned? i tried but didn’t get much result

ls -a in Linux will reveal all the files in the current directory including the hidden ones…
There might be some files which are causing the problem…

(Just as Rik mentioned)

Try this code (looks promising)

hi Aditya, you were right that the .ipynb checkpoints is messing it up. How should I remove it?

Just remove them…
Either from the notebook(delete) or use the terminal…

it is hidden so how do i exactly remove them from the notebook?

Oh I see You can go to that particular directory from the terminal (Linux) and use the rm command…
(Search for it on Internet, I forgot exactly…
Most probably it’s rm -rf …
But have a look)

Cheers mate I finally removed all of the ipynb_checkpoints. it was a real pain there were quite a number of them in different folders.

for those who have similar errors, simply run (in jupyter):

!cd {PATH} && rm -rf .ipynb_checkpoints
(remove ‘!’ if running from terminal)

Shoutout to ecdrid for all the help

Glad that you have made it…

But a humble request will be to first use the search bar in the forum…

Most of the issues were answered by folks…(Scrolling till the end)


Yes, this issue has already been addressed in the beginner’s forum here:
How to remove .ipynb checkpoint