数据增加后数据集尺寸

Question

在我使用数据增强之后，类似

train = ImageDataGenerator(
        rotation_range=90,    
        width_shift_range=0.2,  
        height_shift_range=0.2,
        zca_whitening = True,
        shear_range=0.2)

test = ImageDataGenerator(rescale=1./255)

train_dataset = train.flow_from_directory("/content/drive/MyDrive/dataset",
                                          target_size=(150, 150),
                                          batch_size = 32,
                                          class_mode = 'binary')

输出：

Found 327 images belonging to 2 classes.

由于327张图像是原始数据大小，在数据增加后我的培训数据集增加了吗？如果是这样，我应该如何获得新的数据集大小？

Answer 1

是的，您可以通过为每个班级创建和保存增强图像，然后将这些图像与原始火车集合并来增加大小。以下是我用于该称为余额的函数。您输入其中的数据框，该列列是文件膜和标签。 Filepath列包含训练图像的完整文件路径。标签是图像的相应类标签。工作目录是存储增强图像的目录。 IMG_SIZE是增强图像的大小。

def balance(df, n, working_dir, img_size):
    def augment(df,n, working_dir, img_size):
        aug_dir=os.path.join(working_dir, 'aug')
        os.mkdir(aug_dir)        
        for label in df['labels'].unique():    
            dir_path=os.path.join(aug_dir,label)    
            os.mkdir(dir_path)
        # create and store the augmented images  
        total=0
        gen=ImageDataGenerator(horizontal_flip=True,  rotation_range=20, width_shift_range=.2,
                                      height_shift_range=.2, zoom_range=.2)
        groups=df.groupby('labels') # group by class
        for label in df['labels'].unique():  # for every class               
            group=groups.get_group(label)  # a dataframe holding only rows with the specified label 
            sample_count=len(group)   # determine how many samples there are in this class  
            if sample_count< n: # if the class has less than target number of images
                aug_img_count=0
                delta=n - sample_count  # number of augmented images to create
                target_dir=os.path.join(aug_dir, label)  # define where to write the images
                msg='{0:40s} for class {1:^30s} creating {2:^5s} augmented images'.format(' ', label, str(delta))
                print(msg, '\r', end='') # prints over on the same line
                aug_gen=gen.flow_from_dataframe( group,  x_col='filepaths', y_col=None, target_size=img_size,
                                                class_mode=None, batch_size=1, shuffle=False, 
                                                save_to_dir=target_dir, save_prefix='aug-', color_mode='rgb',
                                                save_format='jpg')
                while aug_img_count<delta:
                    images=next(aug_gen)            
                    aug_img_count += len(images)
                total +=aug_img_count
        print('Total Augmented images created= ', total)
        # create aug_df and merge with train_df to create composite training set ndf
        aug_fpaths=[]
        aug_labels=[]
        classlist=os.listdir(aug_dir)
        for klass in classlist:
            classpath=os.path.join(aug_dir, klass)     
            flist=os.listdir(classpath)    
            for f in flist:        
                fpath=os.path.join(classpath,f)         
                aug_fpaths.append(fpath)
                aug_labels.append(klass)
        Fseries=pd.Series(aug_fpaths, name='filepaths')
        Lseries=pd.Series(aug_labels, name='labels')
        aug_df=pd.concat([Fseries, Lseries], axis=1)         
        df=pd.concat([df,aug_df], axis=0).reset_index(drop=True)        
        return df 
    
    df=df.copy() 
    # make directories to store augmented images
    aug_dir=os.path.join(working_dir, 'aug')    
    if 'aug' in os.listdir(working_dir):
        print(' Augmented images already exist. To delete these and create new images enter D, else enter U to use these images', flush=True)
        ans=input(' ')
        if ans == 'D' or ans == 'd':            
            shutil.rmtree(aug_dir) # start with an clean empty directory  
            return augment(df,n, working_dir, img_size)
            
        else:
            return df
    else:
        return augment(df,n, working_dir, img_size)
        
        
   
n=120# number of samples in each class
working_dir=r'./' # directory to store augmented images
img_size=(166,208) # size of augmented images
train_df=balance(train_df, n, working_dir, img_size)