wwwwwwwwwwwwwwww

# Copyright (c) Horizon Robotics. All rights reserved. # Source code reference to mmdetection, gluoncv, gluonnas. import copy import inspect import logging import math import random from abc import abstractmethod from typing import Dict, List, Optional, Sequence, Tuple, Union import cv2 import numpy as np import torch import torchvision from hat.core.box_utils import bbox_overlaps, is_center_of_bboxes_in_roi from hat.data.transforms.functional_img import ( image_normalize, image_pad, imresize, imresize_pad_to_keep_ratio, random_flip, ) from hat.registry import OBJECT_REGISTRY from .affine import ( AffineAugMat, AffineMat2DGenerator, AffineMatFromROIBoxGenerator, AlphaImagePyramid, ImageAffineTransform, LabelAffineTransform, _pad_array, get_affine_image_resize, resize_affine_mat, ) from .bbox import ( clip_bbox, filter_bbox, remap_bbox_label_by_area, remap_bbox_label_by_clip_area_ratio, ) from .classification import BgrToYuv444, BgrToYuv444V2 from .common import Cast from .gaze import eye_ldmk_mirror try: import albumentations except ImportError: albumentations = None logger = logging.getLogger(__name__) __all__ = [ "Resize", "Resize3D", "RandomFlip", "Pad", "Normalize", "RandomCrop", "ToTensor", "Batchify", "FixedCrop", "PresetCrop", "ColorJitter", "RandomExpand", "MinIoURandomCrop", "AugmentHSV", "get_dynamic_roi_from_camera", "IterableDetRoITransform", "PadDetData", "ToFasterRCNNData", "ToMultiTaskFasterRCNNData", "PadTensorListToBatch", "PlainCopyPaste", "HueSaturationValue", "RGBShift", "MeanBlur", "MedianBlur", "RandomBrightnessContrast", "ShiftScaleRotate", "RandomResizedCrop", "AlbuImageOnlyTransform", "BoxJitter", "RandomSizeCrop", "DetYOLOv5MixUp", "DetYOLOXMixUp", "DetMosaic", "DetAffineAugTransformer", "IterableDetRoIListTransform", ] def _transform_bboxes( gt_boxes, ig_regions, img_roi, affine_aug_param, clip=True, min_valid_area=8, min_valid_clip_area_ratio=0.5, min_edge_size=2, complete_boxes=False, ): bbox_ts = LabelAffineTransform(label_type="box") ts_gt_boxes = bbox_ts( gt_boxes, affine_aug_param.mat, flip=affine_aug_param.flipped ) if clip: clip_gt_boxes = clip_bbox(ts_gt_boxes, img_roi, need_copy=True) else: clip_gt_boxes = ts_gt_boxes clip_gt_boxes = remap_bbox_label_by_area(clip_gt_boxes, min_valid_area) clip_gt_boxes = remap_bbox_label_by_clip_area_ratio( ts_gt_boxes, clip_gt_boxes, min_valid_clip_area_ratio ) if clip and complete_boxes: mask = filter_bbox( clip_gt_boxes, img_roi, allow_outside_center=True, min_edge_size=min_edge_size, return_mask=True, ) to_be_hard_flag = np.logical_and( clip_gt_boxes[:, 4] < 0, ts_gt_boxes[:, 4] > 0 ) ts_gt_boxes[to_be_hard_flag, 4] *= -1 clip_gt_boxes = ts_gt_boxes[mask] else: clip_gt_boxes = filter_bbox( clip_gt_boxes, img_roi, allow_outside_center=True, min_edge_size=min_edge_size, ) if ig_regions is not None: ts_ig_regions = bbox_ts( ig_regions, affine_aug_param.mat, flip=affine_aug_param.flipped ) if clip: clip_ig_regions = clip_bbox(ts_ig_regions, img_roi) else: clip_ig_regions = ts_ig_regions else: clip_ig_regions = None return clip_gt_boxes, clip_ig_regions @OBJECT_REGISTRY.register class Resize(object): """Resize image & bbox & mask & seg. .. note:: Affected keys: 'img', 'ori_img', 'img_shape', 'pad_shape', 'resized_shape', 'pad_shape', 'scale_factor', 'gt_bboxes', 'gt_seg', 'gt_ldmk'. Args: img_scale: See above. max_scale: The max size of image. If the image's shape > max_scale, The image is resized to max_scale multiscale_mode: Value must be one of "range" or "value". This transform resizes the input image and bbox to same scale factor. There are 3 multiscale modes: 'ratio_range' is not None: randomly sample a ratio from the ratio range and multiply with the image scale. e.g. Resize(img_scale=(400, 500)), multiscale_mode='range', ratio_range=(0.5, 2.0) 'ratio_range' is None and 'multiscale_mode' == "range": randomly sample a scale from a range, the length of img_scale[tuple] must be 2, which represent small img_scale and large img_scale. e.g. Resize(img_scale=((100, 200), (400,500)), multiscale_mode='range') 'ratio_range' is None and 'multiscale_mode' == "value": randomly sample a scale from multiple scales. e.g. Resize(img_scale=((100, 200), (300, 400), (400, 500)), multiscale_mode='value'))) ratio_range: Scale factor range like (min_ratio, max_ratio). keep_ratio: Whether to keep the aspect ratio when resizing the image. pad_to_keep_ratio: Whether to pad image to keep the same shape and aspect ratio when resizing the image to target shape. raw_scaler_enable: Whether to enable raw scaler when resize the image. sample1c_enable: Whether to sample one channel after resize the image. divisor: Width and height are rounded to multiples of `divisor`. rm_neg_coords: Whether to rm negative coordinates. """ def __init__( self, img_scale: Union[Sequence[int], Sequence[Sequence[int]]] = None, max_scale: Union[Sequence[int], Sequence[Sequence[int]]] = None, multiscale_mode: str = "range", ratio_range: Tuple[float, float] = None, keep_ratio: bool = True, pad_to_keep_ratio: bool = False, raw_scaler_enable: bool = False, sample1c_enable: bool = False, divisor: int = 1, rm_neg_coords: bool = True, split_transform: bool = False, split_trans_w: int = 256, split_trans_h: int = 256, ): if img_scale is None: self.img_scale = img_scale else: if isinstance(img_scale, (list, tuple)): if isinstance(img_scale[0], (tuple, list)): self.img_scale = img_scale else: self.img_scale = [img_scale] else: self.img_scale = [img_scale] for value in self.img_scale: assert isinstance(value, (tuple, list)), ( "you should set img_scale like a tupe/list or a list of " "tuple/list" ) self.max_scale = max_scale if ratio_range is not None: # mode 1: given a scale and a range of image ratio assert len(self.img_scale) == 1 else: # mode 2: given multiple scales or a range of scales assert multiscale_mode in ["value", "range", "max_size"] self.multiscale_mode = multiscale_mode self.ratio_range = ratio_range self.keep_ratio = keep_ratio self.pad_to_keep_ratio = pad_to_keep_ratio self.raw_scaler_enable = raw_scaler_enable self.sample1c_enable = sample1c_enable self.divisor = divisor self.rm_neg_coords = rm_neg_coords self.split_transform = split_transform self.split_trans_w = split_trans_w self.split_trans_h = split_trans_h @staticmethod def random_select(img_scales): scale_idx = np.random.randint(len(img_scales)) img_scale = img_scales[scale_idx] return img_scale, scale_idx @staticmethod def random_sample(img_scales): assert len(img_scales) == 2