lc82111 / keras_hed Goto Github PK

What is the necessary configuration needed?

I have trained the HED-BSDS

and , how can I test the dataset?

I can't find the code about testing.

thx

Hello, I tried your code with the HED dataset. When i run main_segmentation.py, It shows Epoch1/4096
and "while True:" for a long time. I think maybe there is somrthing wrong with the generate_minibatches().
Do you have the same problem?

Hello~
I have trained and tested this HED model.
In conclusion, I have confirmed this model works fine and very good.
I saved the model as .h5 format using model.save('hed.h5').
Then, I have tried to load the saved model again using keras.models.load_model('hed.h5').
But the error occurred like:
ValueError: Unknown loss function:cross_entropy_balanced

How can I get over this?

(detailed error message)
Traceback (most recent call last):

File "", line 1, in
model0 = keras.models.load_model('hed.h5')

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\engine\saving.py", line 492, in load_wrapper
return load_function(*args, **kwargs)

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\engine\saving.py", line 584, in load_model
model = _deserialize_model(h5dict, custom_objects, compile)

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\engine\saving.py", line 369, in _deserialize_model
sample_weight_mode=sample_weight_mode)

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\engine\training.py", line 119, in compile
self.loss, self.output_names)

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\engine\training_utils.py", line 820, in prepare_loss_functions
loss_functions.append(get_loss_function(loss.get(name, None)))

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\engine\training_utils.py", line 705, in get_loss_function
loss_fn = losses.get(loss)

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\losses.py", line 795, in get
return deserialize(identifier)

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\losses.py", line 776, in deserialize
printable_module_name='loss function')

File "C:\Anaconda3\lib\site-packages\keras-2.3.0-py3.6.egg\keras\utils\generic_utils.py", line 167, in deserialize_keras_object
':' + function_name)

ValueError: Unknown loss function:cross_entropy_balanced

I trained this model, but on testing I observe that est[4][0] looks very weird.
Below are the outputs from the model : gt, then o1-o5 and fuse.
All outputs are good except o5. Does anyone have this issue?

Hi! Can you Provide pre trained weights for your Model? That would be a Great Advantage as Starting Point. Thanks.

Hi~ I have tested this well. I think this is very good example and I will try to enhance this model.
But, I have a question.
Why are the correction values of 103.939, 116.779, and 123.68 to BGR channels used in HED_data_parser.py file?
Why did you specify these values?

Hi,
thanks for your work.
have you tested the model accuracy ODS, OIS ?
and how many epoch does it need?
Regards