CODEWITHSUNDEEP
pythontensorflowkerasconv-neural-networkvalueerror
J. Dav
J. Dav

Reputation: 31

ValueError: Input 0 is incompatible with layer conv2d_1: expected ndim=4, found ndim=3

After inquiring into the questions already asked about this problem, I keep presenting it. Im trying to classify letters from A to D. All input images are 64x64 and graycolor.

The first layer of my CNN is:

model = Sequential()
model.add(Conv2D(32, (3,  3), input_shape = input_shape, activation = 'relu'))

And input_shape it's coming from:

# Define the number of classes
num_classes = 4
labels_name={'A':0,'B':1,'C':2,'D':3}

img_data_list=[]
labels_list=[]

for dataset in data_dir_list:
    img_list=os.listdir(data_path+'/'+ dataset)
    print ('Loading the images of dataset-'+'{}\n'.format(dataset))
    label = labels_name[dataset]
    for img in img_list:
    input_img=cv2.imread(data_path + '/'+ dataset + '/'+ img )
          input_img=cv2.cvtColor(input_img, cv2.COLOR_BGR2GRAY)
          input_img_resize=cv2.resize(input_img,(128,128))
          img_data_list.append(input_img_resize)
          labels_list.append(label)

img_data = np.array(img_data_list)
img_data = img_data.astype('float32')
img_data /= 255
print (img_data.shape)

labels = np.array(labels_list)
print(np.unique(labels,return_counts=True))

#convert class labels to on-hot encoding
Y = np_utils.to_categorical(labels, num_classes)

#Shuffle the dataset
x,y = shuffle(img_data,Y, random_state=2)

# Split the dataset
X_train, X_test, y_train, y_test = train_test_split(x, y, test_size=0.2, random_state=2)

#Defining the model
input_shape=img_data[0].shape
print(input_shape)

Upvotes: 2

Views: 8352

Answers (2)

Rakesh Dhanekula
Rakesh Dhanekula

Reputation: 21

The CNN model needs a data set with dimensions greater when used multiple layer(Convolutional layers and pooling). To avoid the negative dimensional problems increase the image dimensions or reduce the CNN layers. It works..

Upvotes: 2

Dinari
Dinari

Reputation: 2557

Conv2d expects input of shape (batchsize, w, h, filters).

You need to add a reshape to fit the data before the conv layer:

 model.add(Reshape((64, 64, 1)))

This will set your model dimensions to [None, 64,64,1] and should be fine for Conv2d.

Upvotes: 4

Related Questions