get_data

hybrid_learning.experimentation.fuzzy_exp.fuzzy_exp_helpers.get_data(img_size, dataset_root, split, root=None, predicate_setts=None, default_person_sizes=None, reduce_gt_masks=None, device=None, show_progress_bars=True, use_boxes=True, _log=None, **_)

Return data triple of COCO keypoint datasets with output transformed to semantic segmentation of persons. Allow to subset the annotations to only contain persons of a certain estimated relative (min, max) size range. Cache subsetted annotations.

Parameters

• img_size (Tuple[int, int]) –

• dataset_root (str) –

• split (str) –

• root (Optional[str]) –

• predicate_setts (Optional[Dict[str, Any]]) –

• default_person_sizes (Optional[Dict[str, Tuple[str, str]]]) –

• reduce_gt_masks (Optional[bool]) –

• device (Optional[str]) –

• show_progress_bars (bool) –

• use_boxes (bool) –

• _log (Optional[Logger]) –

Return type

KeypointsDataset