Is it necessary that input to VGG16 should be a 4D Array? I tried passing normal 3-D array. It did not worked. After expanding the array using np.expand_dim() then it worked for me
            Asked
            
        
        
            Active
            
        
            Viewed 80 times
        
    1 Answers
0
            
            
        Yes, CNNs in keras take as input : (batch_size, img_size1, img_size2, channels)
For more information you can check "The input shape" chapter of this answer : https://stackoverflow.com/a/44748370/12384070
 
    
    
        SashimiDélicieux
        
- 476
- 1
- 5
- 17
