search

idealo / image-super-resolution

🔎 Super-scale your images and run experiments with Residual Dense and Adversarial Networks.
https://idealo.github.io/image-super-resolution/
Apache License 2.0
4.6k stars 758 forks source link

Training Deprication Error for creating ndarray out of ragged nested sequences results in failure #180

Open sephethus opened 3 years ago

sephethus commented 3 years ago 
~/image-super-resolution/ISR/utils/datahandler.py:125: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  crops['hr'] = np.array(crops['hr'])
~/image-super-resolution/ISR/utils/datahandler.py:154: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray

File "~/image-super-resolution/ISR/train/trainer.py", line 330, in train
    d_loss_real = self.discriminator.model.train_on_batch(batch['hr'], valid)
ValueError: Failed to convert a NumPy array to a Tensor (Unsupported object type Array)

This happened on epoch 53 out of 100 which is odd. Anybody else getting this error? It's tough because I have a large/growing dataset of 3000 photos and it fails while I'm off doing other things.

zeqiong-06 commented 2 years ago

Have you solved this issue?