Hi, is it possible to train the model on a grayscale dataset without transforming it to rgb? I tried the example code with c_dim=1 in the RRDN model but I have an error with the dimensions of the next layers:
WARNING:tensorflow:Model was constructed with shape (None, 80, 80, 3) for input KerasTensor(type_spec=TensorSpec(shape=(None, 80, 80, 3), dtype=tf.float32, name='input_2'), name='input_2', description="created by layer 'input_2'"), but it was called on an input with incompatible shape (None, 80, 80, 1).
Traceback (most recent call last):
File "/home/dolabok/Documents/[...]/super_resolution.py", line 76, in
trainer = Trainer(
File "/home/dolabok/anaconda3/lib/python3.9/site-packages/ISR-2.2.0-py3.9.egg/ISR/train/trainer.py", line 105, in init
File "/home/dolabok/anaconda3/lib/python3.9/site-packages/ISR-2.2.0-py3.9.egg/ISR/train/trainer.py", line 185, in _combine_networks
File "/home/dolabok/anaconda3/lib/python3.9/site-packages/keras/utils/traceback_utils.py", line 70, in error_handler
raise e.with_traceback(filtered_tb) from None
File "/home/dolabok/anaconda3/lib/python3.9/site-packages/keras/engine/input_spec.py", line 277, in assert_input_compatibility
raise ValueError(
ValueError: Exception encountered when calling layer "discriminator" (type Functional).
Input 0 of layer "Conv_1" is incompatible with the layer: expected axis -1 of input shape to have value 3, but received input with shape (None, 80, 80, 1)
Call arguments received by layer "discriminator" (type Functional):
• inputs=tf.Tensor(shape=(None, 80, 80, 1), dtype=float32)
• training=False
• mask=None
it seems that the discriminator is hard-coded with channel = 3
Hi, is it possible to train the model on a grayscale dataset without transforming it to rgb? I tried the example code with c_dim=1 in the RRDN model but I have an error with the dimensions of the next layers:
WARNING:tensorflow:Model was constructed with shape (None, 80, 80, 3) for input KerasTensor(type_spec=TensorSpec(shape=(None, 80, 80, 3), dtype=tf.float32, name='input_2'), name='input_2', description="created by layer 'input_2'"), but it was called on an input with incompatible shape (None, 80, 80, 1). Traceback (most recent call last): File "/home/dolabok/Documents/[...]/super_resolution.py", line 76, in
trainer = Trainer(
File "/home/dolabok/anaconda3/lib/python3.9/site-packages/ISR-2.2.0-py3.9.egg/ISR/train/trainer.py", line 105, in init
File "/home/dolabok/anaconda3/lib/python3.9/site-packages/ISR-2.2.0-py3.9.egg/ISR/train/trainer.py", line 185, in _combine_networks
File "/home/dolabok/anaconda3/lib/python3.9/site-packages/keras/utils/traceback_utils.py", line 70, in error_handler
raise e.with_traceback(filtered_tb) from None
File "/home/dolabok/anaconda3/lib/python3.9/site-packages/keras/engine/input_spec.py", line 277, in assert_input_compatibility
raise ValueError(
ValueError: Exception encountered when calling layer "discriminator" (type Functional).
Input 0 of layer "Conv_1" is incompatible with the layer: expected axis -1 of input shape to have value 3, but received input with shape (None, 80, 80, 1)
Call arguments received by layer "discriminator" (type Functional): • inputs=tf.Tensor(shape=(None, 80, 80, 1), dtype=float32) • training=False • mask=None
it seems that the discriminator is hard-coded with channel = 3
My code :