dhkim0225 / keras-image-segmentation Goto Github PK

Just added a line in test.py after line 57 👍

x_img =cv2.resize(x_img,(512,256))

hi,i have a question,why steps_per_epoch=3475 while not steps_per_epoch=2975 ？the number of the train datasets of cityscapes is 2975，Thank you very much and look forward to your reply

May in know the format for dataset?

Traceback (most recent call last):
File "train.py", line 12, in
from dataset_parser.generator import data_generator
ImportError: No module named dataset_parser.generator

What error is it?How can I solve it?

Traceback (most recent call last):
File "train.py", line 42, in
lr_init=lr_init, lr_decay=lr_decay, vgg_weight_path=vgg_path)
File "/home/atsushi/keras/work/keras-image-segmentation/model/fcn.py", line 111, in fcn_8s
x = Lambda(lambda x: tf.image.resize_images(x, (x.shape[1] * 2, x.shape[2] * 2)))(x)
File "/home/atsushi/.local/lib/python2.7/site-packages/keras/engine/topology.py", line 617, in call
output = self.call(inputs, **kwargs)
File "/home/atsushi/.local/lib/python2.7/site-packages/keras/layers/core.py", line 663, in call
return self.function(inputs, **arguments)
File "/home/atsushi/keras/work/keras-image-segmentation/model/fcn.py", line 111, in
x = Lambda(lambda x: tf.image.resize_images(x, (x.shape[1] * 2, x.shape[2] * 2)))(x)
File "/home/atsushi/tensorflow/local/lib/python2.7/site-packages/tensorflow/python/ops/image_ops_impl.py", line 746, in resize_images
raise ValueError(''size' must be a 1-D int32 Tensor')
ValueError: 'size' must be a 1-D int32 Tensor