class weights using customized generator function keras

Question

I have written the following generator function. In order to use class weights, with this generator, I receive an error when I use the following commands

training_generator=image_generator(partition['train'], labels, bat_siz )

counter = Counter(training_generator.classes) Error: AttributeError: 'generator' object has no attribute 'classes'

Generator function

def image_generator(list_IDs, label_file, batch_size ):
    a=np.arange(0,len(list_IDs),1)   
    a_hat=a
    while True:
          
          batch_path=np.random.choice(a_hat,size = batch_size,replace=False)                                            
          list_IDs_temp = [list_IDs[k] for k in batch_path]
          batch_input  = []
          batch_output = [] 
          
          # Read in each input, perform preprocessing and get labels
          for i, ID in enumerate(list_IDs_temp):
              inp = get_input(ID )
              out_label = get_output(ID,label_file )
              batch_input += [ inp ]
              batch_output += [ out_label ]
              
          # Return a tuple of (input, output) to feed the network
          batch_x = np.array( batch_input )
          batch_y = np.array( batch_output )
          
          rem_sample=np.setdiff1d(a_hat,batch_path) 
          a_hat=rem_sample
          if len(rem_sample)<1:
              a_hat=a
          del  batch_path,list_IDs_temp                 
        
          yield( np.asarray(batch_x), to_categorical(batch_y, num_classes=2) )

Could someone tell how to deal with customized generator function. Thanks in advance

class weights using customized generator function keras

0 Answers0