def process_image()

in PyTorchClassification/data_loader.py [0:0]

• 50 lines of code
• 10 McCabe index (conditional complexity)

    def process_image(self, img, is_train, multi_crop = False, bboxes = None, no_crop=False):
        '''
        Pre-processing of the images
        Arguments:
            img:                    single input image (PIL)
            is_train:               True for training mode, false for validation/testing mode
            multi_crop (optional):  If True, uses 12 crops in validation
            bboxes (optional):      Bounding boxes of the foreground object
            no_crop (optional):     If True, skips cropping in in both training and validation
        '''
        if bboxes is None:
            bboxes = []
        # In training, random scaling, flipping, and color augmentation
        if is_train:
            if no_crop:
                img = self.resize(img)
            else:
                img = self.scale_aug(img)
            img = self.flip_aug(img)
            img = self.color_aug(img)
            img = self.tensor_aug(img)
            img = self.norm_aug(img)
            return img
        # In validation
        else:
            # We will collect all crops of the image in *imgs*
            if no_crop:
                imgs = [self.resize(img)]
            else:
                min_size = min(img.size)
                scale_ratio = min(self.im_size) / min_size * 1.3
                resized_img = F.resize(img, (int(img.size[1]*scale_ratio), int(img.size[0]*scale_ratio)))
                imgs = [self.center_crop(resized_img)]

            # Add all bboxes and their flip
            for bbox in bboxes:
                bbox_shape = np.array([bbox[2] - bbox[0], bbox[3] - bbox[1]])
                padding = bbox_shape.max() * 0.1
                # Add offset to the shorter side to crop a square patch
                offset = (bbox_shape - np.min(bbox_shape))[::-1] // 2
                bbox_crop = img.crop((bbox[1] - padding - offset[1], 
                           bbox[0] - padding - offset[0],
                           bbox[3] + padding + offset[1], 
                           bbox[2] + padding + offset[0])) # (w - crop_w, h - crop_h, w, h))
                #img.save('crop{}.jpg'.format(np.random.randint(0,10)))
                bbox_crop = self.resize(bbox_crop)
                imgs.append(bbox_crop)
                imgs.append(self.flip(bbox_crop))

            # Add all crops 
            if multi_crop:
                imgs.append(self.flip(self.center_crop(resized_img)))
                imgs.extend(self.multi_crop(self.resize_for_crop(img)))

            # Convert everything to normalized tensor
            tensor_imgs = []
            for img in imgs:
                img = self.tensor_aug(img)
                img = self.norm_aug(img)
                tensor_imgs.append(img)
            return tensor_imgs