CenterCrop¶

class mmpretrain.datasets.transforms.CenterCrop(crop_size, auto_pad=False, pad_cfg={'type': 'Pad'}, clip_object_border=True)[source]¶

Crop the center of the image, segmentation masks, bounding boxes and key points. If the crop area exceeds the original image and auto_pad is True, the original image will be padded before cropping.

Required Keys:

img
gt_seg_map (optional)
gt_bboxes (optional)
gt_keypoints (optional)

Modified Keys:

img
img_shape
gt_seg_map (optional)
gt_bboxes (optional)
gt_keypoints (optional)

Added Key:

pad_shape

Parameters:

crop_size (Union[int, Tuple[int, int]]) – Expected size after cropping with the format of (w, h). If set to an integer, then cropping width and height are equal to this integer.
auto_pad (bool) – Whether to pad the image if it’s smaller than the crop_size. Defaults to False.
pad_cfg (dict) – Base config for padding. Refer to mmcv.Pad for detail. Defaults to dict(type='Pad').
clip_object_border (bool) – Whether to clip the objects outside the border of the image. In some dataset like MOT17, the gt bboxes are allowed to cross the border of images. Therefore, we don’t need to clip the gt bboxes in these cases. Defaults to True.

transform(results)[source]¶

Apply center crop on results.

Parameters:: results (dict) – Result dict contains the data to transform.
Returns:: Results with CenterCropped image and semantic segmentation map.
Return type:: dict