taspinar/sidl

Deep Learning code

sidl's Issues

Error when running CNN (Step 6)

I get a matrix dimension error when running the CNN:
InvalidArgumentError: Dimensions must be equal, but are 32768 and 25088 for 'MatMul' (op: 'MatMul') with input shapes: [16,32768], [25088,4096].

It looks like this error involves Variable_26:0 in the vggnet16?

index 5655 is out of bounds for axis 1 with size 0

hi.
I read your blog post and its really amazing for me.
I have an error in section 4.1 Preparing a training, test and validation dataset.
and as you see my stats bellow is different with your in notenook:

There are 4069 tiles containing

this is my outputs:

There are 5701 tiles containing roads.
Of roadtype P (Provincie) there are 47 tiles.
Of roadtype G (Gemeente) there are 4946 tiles.
Of roadtype W (Waterschap) there are 902 tiles.
Of roadtype T (Andere wegbeheerder) there are 1 tiles.
Randomizing dataset...
Traceback (most recent call last):
  File "stats.py", line 227, in <module>
    dataset, labels_roadpresence, labels_roadtype_ohe, labels_filename = reformat_data(dataset, labels_roadpresence, labels_roadtype_ohe, labels_filename)
  File "stats.py", line 184, in reformat_data
    dataset, labels1, labels2, labels3 = randomize(dataset, labels1, labels2, labels3)
  File "stats.py", line 176, in randomize
    randomized_labels2 = labels2[permutation]
IndexError: index 5655 is out of bounds for axis 1 with size 0

the lines error log pointed is:

def randomize(dataset, labels1, labels2, labels3):
    permutation = np.random.permutation(dataset.shape[0])
    randomized_dataset = dataset[permutation, :, :, :]
    randomized_labels1 = labels1[permutation]
    randomized_labels2 = labels2[permutation]
    randomized_labels3 = labels3[permutation]

and
dataset, labels_roadpresence, labels_roadtype_ohe, labels_filename = reformat_data(dataset, labels_roadpresence, labels_roadtype_ohe, labels_filename)

what can I do for next chapter?

Needed help with debugging the errors in code

There are several errors arising while I was trying to run the code.
like:
'utf-8' codec can't decode byte 0xfc in position 1: invalid start byte
Please debug the code once by today... It will be real help!

Loading MNIST Dataset

Running "load_data.py" gives error on "from mnist import MNIST".
Can you please clarify as to from where is this imported.

File names get out of synch

Hi, firstly thanks v much for the work. Im new to python and this so its been great. I have a problem in that the download works perfect and I create the Jason file but when it goes through trying to read the files it comes up with filenames that dont exist in the folder...Its as if the function that is coming up with the names has a bigger range than the actual existing files...
Do you have a complete script?
Many thanks again
PS the skipy imread is also now deprecated I think

ValueError: Dimensions must be equal, but are 32768 and 25088 for ‘MatMul’ (op: ‘MatMul’) with input shapes: [16,32768], [25088,4096]

Hi,

I encounter this error: ValueError: Dimensions must be equal, but are 32768 and 25088 for ‘MatMul’ (op: ‘MatMul’) with input shapes: [16,32768], [25088,4096]

How could I solve that? Any idea?