generator.py

class GetDataset():
    """Claim Dataset object for inferencing.
    
    Args:
        slide_name (str): full path to slide
        f_inputs_preproc (function): preprocessing function to array
        patch_size (int): patch size of array
        stride (int): stride of image when get patches
        level (int): get patch at level
    
    Returns:
        Array
    """
    def __init__(self, slide_name, f_inputs_preproc, patch_size=256, stride=256, level=0):
        self.slide_name = slide_name
        self.preproc = f_inputs_preproc
        self.patch_size = patch_size
        self.level = level
        
        ## Init ##
        self.this_slide = Slide_reader(slide_name=slide_name, rle_file=None, show_info=False)
        slide_w, slide_h = self.this_slide.slide_info["Width"], self.this_slide.slide_info['Height']
        self.croplist = self.compute_croplist(width=slide_w, height=slide_h, stride=stride)
        self.lowRes_w, self.lowRes_h = slide_w//patch_size, slide_h//patch_size
        self.counter = 0
    
    def __len__(self):
        return len(self.croplist)
    
    def __getitem__(self, idx):
        this_coord = self.croplist[idx]
        arr = self.this_slide.get_patch_at_level(coord=this_coord, 
                                                 sz=(self.patch_size, self.patch_size), 
                                                 level=self.level)
        if self.preproc is not None:
            arr = self.preproc(arr)
            
        self.counter = (self.counter + 1) % len(self.croplist)
        return arr
    
    def __next__(self):
        return self.__getitem__(idx=self.counter)
    
    @staticmethod
    def compute_croplist(width, height, stride):
        x_list = range(0, width, stride)
        y_list = range(0, height, stride)
        
        return list(itertools.product(x_list, y_list))
        
        
class Inference_Dataloader(tf.keras.utils.Sequence):
    """Compose multiple generators together
    Args:
        - slide_obj (class): 
        - batch_size (int):
    Yield:
        - Batch of data
    """
    def __init__(self, slide_object, batch_size):
        self.slide_object = slide_object
        self.batch_size = batch_size
        
    def __len__(self):
        return math.ceil(len(self.slide_object) / self.batch_size)
        
    def __getitem__(self, idx):
        idx_start = idx * self.batch_size
        idx_end = (idx+1) * self.batch_size
        
        imgs = []
        for j in range(idx_start, idx_end):
            img = self.slide_object[j]
            imgs.append(img)        
        return np.array(imgs)