hasnainraz / fc-densenet-tensorflow Goto Github PK

I wonder why set 'training' True in function infer in the model.py

Hi, thanks for your sharing. I learned a lot!
I am little confused for the training mode in the Model.py (line#: 339). To get the dice for the validation set, we should change the training mode from 'True' to 'False', is that right? Because we we have trained the weights by the training set, and we should not train any weights by the validation set.

I am a new learner. I will be very grateful if you reply me.
Thanks a lot!

Thanks for your work. I have a question that after training how to test the model on my own data sets?

python main.py --mode=train
Traceback (most recent call last):
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py", line 510, in _apply_op_helper
preferred_dtype=default_dtype)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 1094, in internal_convert_to_tensor
ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 931, in _TensorTensorConversionFunction
(dtype.name, t.dtype.name, str(t)))

ValueError: Tensor conversion requested dtype string for Tensor with dtype float32: 'Tensor("arg0:0", shape=(), dtype=float32)'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "main.py", line 44, in
main()
File "main.py", line 38, in main
FLAGS.batch_size, FLAGS.epochs, FLAGS.learning_rate)
File "/home/user10/notebook/tiramisu/model.py", line 281, in train
train_image_paths, train_mask_paths, batch_size)
File "/home/user10/notebook/tiramisu/utility.py", line 86, in data_batch
num_parallel_calls=num_threads).prefetch(30)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/data/ops/dataset_ops.py", line 988, in map
return ParallelMapDataset(self, map_func, num_parallel_calls)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/data/ops/dataset_ops.py", line 2230, in init
super(ParallelMapDataset, self).init(input_dataset, map_func)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/data/ops/dataset_ops.py", line 2198, in init
map_func, "Dataset.map()", input_dataset)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/data/ops/dataset_ops.py", line 1454, in init
self._function.add_to_graph(ops.get_default_graph())
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/function.py", line 481, in add_to_graph
self._create_definition_if_needed()
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/function.py", line 337, in _create_definition_if_needed
self._create_definition_if_needed_impl()
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/function.py", line 346, in _create_definition_if_needed_impl
self._capture_by_value, self._caller_device)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/function.py", line 863, in func_graph_from_py_func
outputs = func(*func_graph.inputs)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/data/ops/dataset_ops.py", line 1392, in tf_data_structured_function_wrapper
ret = func(*nested_args)
File "/home/user10/notebook/tiramisu/utility.py", line 5, in _parse_data_infer
image_content = tf.read_file(image_paths)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/ops/gen_io_ops.py", line 528, in read_file
"ReadFile", filename=filename, name=name)
File "/home/user10/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py", line 533, in _apply_op_helper
(prefix, dtypes.as_dtype(input_arg.type).name))

TypeError: Input 'filename' of 'ReadFile' Op has type float32 that does not match expected type of string.

before training, why should image/255 and mask/255 my mask's each pixel is an index in [0,classnumber] if mask/255 they will all be 0 (class:background) when i use pascal VOC 2012 dataset
one more thing why use 30 in parse resize normaliza and shuffle thanks a lot

i didnt change any word in model.py

Traceback (most recent call last):
File "main.py", line 44, in
main()
File "main.py", line 38, in main
FLAGS.batch_size, FLAGS.epochs, FLAGS.learning_rate)
File "/home/user/CHI_FCN/FC-DenseNet-TensorFlow-master/model.py", line 347, in train
total_train_cost / train_step, total_val_cost / val_step))
UnboundLocalError: local variable 'train_step' referenced before assignment

after training,model will save three ckpt files, .index .meta and .data, how can I run infer?I run the
python main.py --ckpt=models/model.ckpt-4
but the pictures in output are all black. And there is no errors and warnings.

In function xentropy_loss() of model.py
logits = tf.reshape(logits, [tf.shape(logits)[0], -1, 2])
should be
logits = tf.reshape(logits, [tf.shape(logits)[0], -1, self.num_classes])

• weight init
  In the original paper, he initialization is used instead of xavier, this would translate to using: tf.contrib.layers.variance_scaling_initializer
• max pooling
  In the original paper, the transition down should be a 2x2 max pool that is non-overlapping. Checking the original Lasagne code it seems they use a 2x2 max pool with a window of 2x2 and stride of 2x2. Wasn't sure if your current 4x4 window as intentional.

Great work on this btw, very readable!

I've set the default Path, in the main.py script, to my Dataset Folder:

parser.add_argument("--train_data", default="./dataset/img", help="Directory for training images")

My folder sctruture inside the project looks like this:

• dataset
  • img
    • images
    • masks

When i call the main.py script without arguments, i get the error: TypeError: Input 'filename' of 'ReadFile' Op has type float32 that does not match expected type of string.

How to set the dataset?

I comment out the normalize function，but the outputs are just a vertical line and blank.

I'm trying to run train & infer.
However I encounter error while infer

InvalidArgumentError (see above for traceback): Assign requires shapes of both tensors to match. lhs shape= [1,1,48,2] rhs shape= [1,1,80,2]
	 [[Node: save/Assign_87 = Assign[T=DT_FLOAT, _class=["loc:@prediction/last_conv1x1/kernel"], use_locking=true, validate_shape=true, _device="/job:localhost/replica:0/task:0/device:CPU:0"](prediction/last_conv1x1/kernel, save/RestoreV2_87)]]

Could you provide some advise, Thanks

I found two issues in the process of running the codes.

1. Model.py (line#: 66), it should be 'def batch_norm(self, x, training, name)'?

2. Model.py (line#: 229), it should be
   'x = self.dense_block(x, training, block_nb - 1, 'up_dense_block_' + str(block_nb))' ?

hey
I am building the FC-dense net.
in the paper. The upsampling path is only using the output of the last dense block
but in your code
print("Building upsample path...")

for i, block_nb in enumerate(range(self.nb_blocks - 1, 0, -1)):
    x = self.transition_up(x, x.get_shape()[-1], 'trans_up_' + str(block_nb))
    x = tf.concat([x, concats[len(concats) - i - 1]], axis=3, name='up_concat_' + str(block_nb))
    print(x.get_shape())
    x = self.dense_block(x, training, block_nb, 'up_dense_block_' + str(block_nb))

here x is the last skip list tensor--->concats[4]
maybe you should write this:
x=dense
before the for loop,so that x will be the output of the dense block
thanks.

when i run the order :
python main.py --mode=train --train_data=/mnt/saners-extend/FC-DenseNet-TensorFlow/data/train --val_data=/mnt/saners-extend/FC-DenseNet-TensorFlow/data/val --layers_per_block=4,5,7,10,12,15 --batch_size=2 --epochs=10 --growth_k=16 --num_classes=2 --learning_rate=0.001

it shows that:
/home/anaconda3/lib/python3.6/site-packages/h5py/init.py:36: FutureWarning: Conversion of the second argument of issubdtype from float to np.floating is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type.
from ._conv import register_converters as _register_converters
First Convolution Out: (?, 256, 256, 48)
Downsample Out: (?, 128, 128, 112)
Downsample Out: (?, 64, 64, 192)
Downsample Out: (?, 32, 32, 304)
Downsample Out: (?, 16, 16, 464)
Downsample Out: (?, 8, 8, 656)
Bottleneck Block: (?, 8, 8, 240)
Upsample after concat: (?, 16, 16, 896)
Upsample after concat: (?, 32, 32, 704)
Upsample after concat: (?, 64, 64, 496)
Upsample after concat: (?, 128, 128, 352)
Upsample after concat: (?, 256, 256, 224)
Mask Prediction: (?, 256, 256, 2)
2018-10-25 15:54:14.641467: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2018-10-25 15:54:14.781306: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:897] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-10-25 15:54:14.781714: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1405] Found device 0 with properties:
name: Tesla P100-PCIE-12GB major: 6 minor: 0 memoryClockRate(GHz): 1.3285
pciBusID: 0000:05:01.0
totalMemory: 11.91GiB freeMemory: 2.58GiB
2018-10-25 15:54:14.781753: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1484] Adding visible gpu devices: 0
2018-10-25 15:54:15.190540: I tensorflow/core/common_runtime/gpu/gpu_device.cc:965] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-10-25 15:54:15.190615: I tensorflow/core/common_runtime/gpu/gpu_device.cc:971] 0
2018-10-25 15:54:15.190628: I tensorflow/core/common_runtime/gpu/gpu_device.cc:984] 0: N
2018-10-25 15:54:15.190885: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1097] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 2277 MB memory) -> physical GPU (device: 0, name: Tesla P100-PCIE-12GB, pci bus id: 0000:05:01.0, compute capability: 6.0)
2018-10-25 15:55:52.442988: E tensorflow/stream_executor/cuda/cuda_dnn.cc:352] Could not create cudnn handle: CUDNN_STATUS_NOT_INITIALIZED
2018-10-25 15:55:52.443130: E tensorflow/stream_executor/cuda/cuda_dnn.cc:360] Possibly insufficient driver version: 384.81.0
Segmentation fault (core dumped)

my environment is cuda9.0, cudnn7.0,tensorflow 1.10.1,anyone can give some advice ? thanks very much

Thank you for your work. I have the problem that the iou always equals to 0.5 when I use my dataset to train the model. I am looking forward to your answers.

Hi, I am interested about how to load the pretrained model parameters of encoder part. Can you give me some suggestion? Thanks!

2*1080ti out of memory ,cityscapesScripts datasets,batch size 4.