I'm writing a simple CNN model to Classification Cat and Dog picture from a local directory named train.

Below are the codes that I have written so far:

import numpy as np
import cv2 as cv
import tensorflow.keras as keras
import os
from sklearn.preprocessing import LabelEncoder
from tensorflow.keras.utils import to_categorical
from tensorflow.keras import layers , models
from sklearn.model_selection import train_test_split

images_vector =[]
images_label =[]

fileNames = os.listdir('train')

for i , f_name in enumerate(fileNames) :

image = cv.imread('train/'   f_name)
    
image = cv.resize(image , (50,50))
image = image/255.0
image = image.flatten()

images_vector.append(image)
images_label.append(f_name.split('.')[0]) 

if i000 == 0 : 
    print(f" [INFO ] : {i} images are processed...")

labelEncoder  = LabelEncoder()
images_label = labelEncoder.fit_transform(images_label)

images_label = to_categorical(images_label)
images_label

X_train , X_test , y_train , y_test = 
train_test_split(images_vector ,images_label  , random_state=40 , train_size=0.8)

print('X_train: '   str(X_train.shape))
print('Y_train: '   str(y_train.shape))
print('X_test:  '    str(X_test.shape))
print('Y_test:  '    str(y_test.shape))

Now after running following code to build model :

net = models.Sequential([
    layers.Conv2D(32 , (3,3) , activation='relu' , input_shape = (1,7500)) ,
    layers.MaxPooling2D(2,2),
    layers.Conv2D(64 , (3,3) , activation='relu'),
    layers.Flatten(),
    layers.Dense(2 , activation='softmax')
])

net.summary()

I got this error :

ValueError: Input 0 of layer "conv2d_96" is incompatible with the layer: expected min_ndim=4, found ndim=3. Full shape received: (None, 1, 7500)

I searched a lot to solving the problem and try and test different shapes but can not find the solution

can any body help me?

CodePudding user response：

The Conv2D layer should be used on a 2-dimensional image with eventually some color channels. To make it work please try the following:

1. To keep 2d image structure, remove:

   image = image.flatten()
   

2. Change input shape for the first Conv2D layer to:

   layers.Conv2D(32, ... input_shape = (50,50,3)