train issue

[xuwanpeng12]

I am currently doing research on depth estimation. When I perform training, there are unsolvable problems. I would like to ask you for advice.

self.semantic_loss = tf.reduce_mean(tf.multiply(tf.nn.sparse_softmax_cross_entropy_with_logits(labels=self.semantic, logits=self.sem_est), self.valid)) * 0.1
File "C:\Users\xwp\Anaconda3\envs\tensorflow\lib\site-packages\tensorflow\python\ops\nn_ops.py", line 2653, in sparse_softmax_cross_entropy_with_logits
logits.get_shape()))
ValueError: Shape mismatch: The shape of labels (received (2, 256, 512, 1)) should equal the shape of logits except for the last dimension (received (1, 2, 256, 512, 34)).

please tell me why, thanks very mnuch!

[Lrq7]

hi, have you solved this? I get the same problem