# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. try: from collections.abc import Sequence except Exception: from collections import Sequence import random import os.path as osp import numpy as np import cv2 from PIL import Image, ImageEnhance from .imgaug_support import execute_imgaug from .ops import * from .box_utils import * import paddlex.utils.logging as logging class DetTransform: """检测数据处理基类 """ def __init__(self): pass class Compose(DetTransform): """根据数据预处理/增强列表对输入数据进行操作。 所有操作的输入图像流形状均是[H, W, C],其中H为图像高,W为图像宽,C为图像通道数。 Args: transforms (list): 数据预处理/增强列表。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 数据长度不匹配。 """ def __init__(self, transforms): if not isinstance(transforms, list): raise TypeError('The transforms must be a list!') if len(transforms) < 1: raise ValueError('The length of transforms ' + \ 'must be equal or larger than 1!') self.transforms = transforms self.use_mixup = False for t in self.transforms: if type(t).__name__ == 'MixupImage': self.use_mixup = True # 检查transforms里面的操作,目前支持PaddleX定义的或者是imgaug操作 for op in self.transforms: if not isinstance(op, DetTransform): import imgaug.augmenters as iaa if not isinstance(op, iaa.Augmenter): raise Exception( "Elements in transforms should be defined in 'paddlex.det.transforms' or class of imgaug.augmenters.Augmenter, see docs here: https://paddlex.readthedocs.io/zh_CN/latest/apis/transforms/" ) def __call__(self, im, im_info=None, label_info=None): """ Args: im (str/np.ndarray): 图像路径/图像np.ndarray数据。 im_info (dict): 存储与图像相关的信息,dict中的字段如下: - im_id (np.ndarray): 图像序列号,形状为(1,)。 - image_shape (np.ndarray): 图像原始大小,形状为(2,), image_shape[0]为高,image_shape[1]为宽。 - mixup (list): list为[im, im_info, label_info],分别对应 与当前图像进行mixup的图像np.ndarray数据、图像相关信息、标注框相关信息; 注意,当前epoch若无需进行mixup,则无该字段。 label_info (dict): 存储与标注框相关的信息,dict中的字段如下: - gt_bbox (np.ndarray): 真实标注框坐标[x1, y1, x2, y2],形状为(n, 4), 其中n代表真实标注框的个数。 - gt_class (np.ndarray): 每个真实标注框对应的类别序号,形状为(n, 1), 其中n代表真实标注框的个数。 - gt_score (np.ndarray): 每个真实标注框对应的混合得分,形状为(n, 1), 其中n代表真实标注框的个数。 - gt_poly (list): 每个真实标注框内的多边形分割区域,每个分割区域由点的x、y坐标组成, 长度为n,其中n代表真实标注框的个数。 - is_crowd (np.ndarray): 每个真实标注框中是否是一组对象,形状为(n, 1), 其中n代表真实标注框的个数。 - difficult (np.ndarray): 每个真实标注框中的对象是否为难识别对象,形状为(n, 1), 其中n代表真实标注框的个数。 Returns: tuple: 根据网络所需字段所组成的tuple; 字段由transforms中的最后一个数据预处理操作决定。 """ def decode_image(im_file, im_info, label_info): if im_info is None: im_info = dict() if isinstance(im_file, np.ndarray): if len(im_file.shape) != 3: raise Exception( "im should be 3-dimensions, but now is {}-dimensions". format(len(im_file.shape))) im = im_file else: try: im = cv2.imread(im_file).astype('float32') except: raise TypeError('Can\'t read The image file {}!'.format( im_file)) im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB) # make default im_info with [h, w, 1] im_info['im_resize_info'] = np.array( [im.shape[0], im.shape[1], 1.], dtype=np.float32) im_info['image_shape'] = np.array([im.shape[0], im.shape[1]]).astype('int32') if not self.use_mixup: if 'mixup' in im_info: del im_info['mixup'] # decode mixup image if 'mixup' in im_info: im_info['mixup'] = \ decode_image(im_info['mixup'][0], im_info['mixup'][1], im_info['mixup'][2]) if label_info is None: return (im, im_info) else: return (im, im_info, label_info) outputs = decode_image(im, im_info, label_info) im = outputs[0] im_info = outputs[1] if len(outputs) == 3: label_info = outputs[2] for op in self.transforms: if im is None: return None if isinstance(op, DetTransform): outputs = op(im, im_info, label_info) im = outputs[0] else: im = execute_imgaug(op, im) if label_info is not None: outputs = (im, im_info, label_info) else: outputs = (im, im_info) return outputs def add_augmenters(self, augmenters): if not isinstance(augmenters, list): raise Exception( "augmenters should be list type in func add_augmenters()") transform_names = [type(x).__name__ for x in self.transforms] for aug in augmenters: if type(aug).__name__ in transform_names: logging.error("{} is already in ComposedTransforms, need to remove it from add_augmenters().".format(type(aug).__name__)) self.transforms = augmenters + self.transforms class ResizeByShort(DetTransform): """根据图像的短边调整图像大小(resize)。 1. 获取图像的长边和短边长度。 2. 根据短边与short_size的比例,计算长边的目标长度, 此时高、宽的resize比例为short_size/原图短边长度。 3. 如果max_size>0,调整resize比例: 如果长边的目标长度>max_size,则高、宽的resize比例为max_size/原图长边长度。 4. 根据调整大小的比例对图像进行resize。 Args: target_size (int): 短边目标长度。默认为800。 max_size (int): 长边目标长度的最大限制。默认为1333。 Raises: TypeError: 形参数据类型不满足需求。 """ def __init__(self, short_size=800, max_size=1333): self.max_size = int(max_size) if not isinstance(short_size, int): raise TypeError( "Type of short_size is invalid. Must be Integer, now is {}". format(type(short_size))) self.short_size = short_size if not (isinstance(self.max_size, int)): raise TypeError("max_size: input type is invalid.") def __call__(self, im, im_info=None, label_info=None): """ Args: im (numnp.ndarraypy): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 其中,im_info更新字段为: - im_resize_info (np.ndarray): resize后的图像高、resize后的图像宽、resize后的图像相对原始图的缩放比例 三者组成的np.ndarray,形状为(3,)。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 数据长度不匹配。 """ if im_info is None: im_info = dict() if not isinstance(im, np.ndarray): raise TypeError("ResizeByShort: image type is not numpy.") if len(im.shape) != 3: raise ValueError('ResizeByShort: image is not 3-dimensional.') im_short_size = min(im.shape[0], im.shape[1]) im_long_size = max(im.shape[0], im.shape[1]) scale = float(self.short_size) / im_short_size if self.max_size > 0 and np.round(scale * im_long_size) > self.max_size: scale = float(self.max_size) / float(im_long_size) resized_width = int(round(im.shape[1] * scale)) resized_height = int(round(im.shape[0] * scale)) im_resize_info = [resized_height, resized_width, scale] im = cv2.resize( im, (resized_width, resized_height), interpolation=cv2.INTER_LINEAR) im_info['im_resize_info'] = np.array(im_resize_info).astype(np.float32) if label_info is None: return (im, im_info) else: return (im, im_info, label_info) class Padding(DetTransform): """1.将图像的长和宽padding至coarsest_stride的倍数。如输入图像为[300, 640], `coarest_stride`为32,则由于300不为32的倍数,因此在图像最右和最下使用0值 进行padding,最终输出图像为[320, 640]。 2.或者,将图像的长和宽padding到target_size指定的shape,如输入的图像为[300,640], a. `target_size` = 960,在图像最右和最下使用0值进行padding,最终输出 图像为[960, 960]。 b. `target_size` = [640, 960],在图像最右和最下使用0值进行padding,最终 输出图像为[640, 960]。 1. 如果coarsest_stride为1,target_size为None则直接返回。 2. 获取图像的高H、宽W。 3. 计算填充后图像的高H_new、宽W_new。 4. 构建大小为(H_new, W_new, 3)像素值为0的np.ndarray, 并将原图的np.ndarray粘贴于左上角。 Args: coarsest_stride (int): 填充后的图像长、宽为该参数的倍数,默认为1。 target_size (int|list|tuple): 填充后的图像长、宽,默认为None,coarset_stride优先级更高。 Raises: TypeError: 形参`target_size`数据类型不满足需求。 ValueError: 形参`target_size`为(list|tuple)时,长度不满足需求。 """ def __init__(self, coarsest_stride=1, target_size=None): self.coarsest_stride = coarsest_stride if target_size is not None: if not isinstance(target_size, int): if not isinstance(target_size, tuple) and not isinstance( target_size, list): raise TypeError( "Padding: Type of target_size must in (int|list|tuple)." ) elif len(target_size) != 2: raise ValueError( "Padding: Length of target_size must equal 2.") self.target_size = target_size def __call__(self, im, im_info=None, label_info=None): """ Args: im (numnp.ndarraypy): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 数据长度不匹配。 ValueError: coarsest_stride,target_size需有且只有一个被指定。 ValueError: target_size小于原图的大小。 """ if im_info is None: im_info = dict() if not isinstance(im, np.ndarray): raise TypeError("Padding: image type is not numpy.") if len(im.shape) != 3: raise ValueError('Padding: image is not 3-dimensional.') im_h, im_w, im_c = im.shape[:] if isinstance(self.target_size, int): padding_im_h = self.target_size padding_im_w = self.target_size elif isinstance(self.target_size, list) or isinstance(self.target_size, tuple): padding_im_w = self.target_size[0] padding_im_h = self.target_size[1] elif self.coarsest_stride > 0: padding_im_h = int( np.ceil(im_h / self.coarsest_stride) * self.coarsest_stride) padding_im_w = int( np.ceil(im_w / self.coarsest_stride) * self.coarsest_stride) else: raise ValueError( "coarsest_stridei(>1) or target_size(list|int) need setting in Padding transform" ) pad_height = padding_im_h - im_h pad_width = padding_im_w - im_w if pad_height < 0 or pad_width < 0: raise ValueError( 'the size of image should be less than target_size, but the size of image ({}, {}), is larger than target_size ({}, {})' .format(im_w, im_h, padding_im_w, padding_im_h)) padding_im = np.zeros( (padding_im_h, padding_im_w, im_c), dtype=np.float32) padding_im[:im_h, :im_w, :] = im if label_info is None: return (padding_im, im_info) else: return (padding_im, im_info, label_info) class Resize(DetTransform): """调整图像大小(resize)。 - 当目标大小(target_size)类型为int时,根据插值方式, 将图像resize为[target_size, target_size]。 - 当目标大小(target_size)类型为list或tuple时,根据插值方式, 将图像resize为target_size。 注意:当插值方式为“RANDOM”时,则随机选取一种插值方式进行resize。 Args: target_size (int/list/tuple): 短边目标长度。默认为608。 interp (str): resize的插值方式,与opencv的插值方式对应,取值范围为 ['NEAREST', 'LINEAR', 'CUBIC', 'AREA', 'LANCZOS4', 'RANDOM']。默认为"LINEAR"。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 插值方式不在['NEAREST', 'LINEAR', 'CUBIC', 'AREA', 'LANCZOS4', 'RANDOM']中。 """ # The interpolation mode interp_dict = { 'NEAREST': cv2.INTER_NEAREST, 'LINEAR': cv2.INTER_LINEAR, 'CUBIC': cv2.INTER_CUBIC, 'AREA': cv2.INTER_AREA, 'LANCZOS4': cv2.INTER_LANCZOS4 } def __init__(self, target_size=608, interp='LINEAR'): self.interp = interp if not (interp == "RANDOM" or interp in self.interp_dict): raise ValueError("interp should be one of {}".format( self.interp_dict.keys())) if isinstance(target_size, list) or isinstance(target_size, tuple): if len(target_size) != 2: raise TypeError( 'when target is list or tuple, it should include 2 elements, but it is {}' .format(target_size)) elif not isinstance(target_size, int): raise TypeError( "Type of target_size is invalid. Must be Integer or List or tuple, now is {}" .format(type(target_size))) self.target_size = target_size def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 数据长度不匹配。 """ if im_info is None: im_info = dict() if not isinstance(im, np.ndarray): raise TypeError("Resize: image type is not numpy.") if len(im.shape) != 3: raise ValueError('Resize: image is not 3-dimensional.') if self.interp == "RANDOM": interp = random.choice(list(self.interp_dict.keys())) else: interp = self.interp im = resize(im, self.target_size, self.interp_dict[interp]) if label_info is None: return (im, im_info) else: return (im, im_info, label_info) class RandomHorizontalFlip(DetTransform): """随机翻转图像、标注框、分割信息,模型训练时的数据增强操作。 1. 随机采样一个0-1之间的小数,当小数小于水平翻转概率时, 执行2-4步操作,否则直接返回。 2. 水平翻转图像。 3. 计算翻转后的真实标注框的坐标,更新label_info中的gt_bbox信息。 4. 计算翻转后的真实分割区域的坐标,更新label_info中的gt_poly信息。 Args: prob (float): 随机水平翻转的概率。默认为0.5。 Raises: TypeError: 形参数据类型不满足需求。 """ def __init__(self, prob=0.5): self.prob = prob if not isinstance(self.prob, float): raise TypeError("RandomHorizontalFlip: input type is invalid.") def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 其中,im_info更新字段为: - gt_bbox (np.ndarray): 水平翻转后的标注框坐标[x1, y1, x2, y2],形状为(n, 4), 其中n代表真实标注框的个数。 - gt_poly (list): 水平翻转后的多边形分割区域的x、y坐标,长度为n, 其中n代表真实标注框的个数。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 数据长度不匹配。 """ if not isinstance(im, np.ndarray): raise TypeError( "RandomHorizontalFlip: image is not a numpy array.") if len(im.shape) != 3: raise ValueError( "RandomHorizontalFlip: image is not 3-dimensional.") if im_info is None or label_info is None: raise TypeError( 'Cannot do RandomHorizontalFlip! ' + 'Becasuse the im_info and label_info can not be None!') if 'gt_bbox' not in label_info: raise TypeError('Cannot do RandomHorizontalFlip! ' + \ 'Becasuse gt_bbox is not in label_info!') image_shape = im_info['image_shape'] gt_bbox = label_info['gt_bbox'] height = image_shape[0] width = image_shape[1] if np.random.uniform(0, 1) < self.prob: im = horizontal_flip(im) if gt_bbox.shape[0] == 0: if label_info is None: return (im, im_info) else: return (im, im_info, label_info) label_info['gt_bbox'] = box_horizontal_flip(gt_bbox, width) if 'gt_poly' in label_info and \ len(label_info['gt_poly']) != 0: label_info['gt_poly'] = segms_horizontal_flip( label_info['gt_poly'], height, width) if label_info is None: return (im, im_info) else: return (im, im_info, label_info) class Normalize(DetTransform): """对图像进行标准化。 1. 归一化图像到到区间[0.0, 1.0]。 2. 对图像进行减均值除以标准差操作。 Args: mean (list): 图像数据集的均值。默认为[0.485, 0.456, 0.406]。 std (list): 图像数据集的标准差。默认为[0.229, 0.224, 0.225]。 Raises: TypeError: 形参数据类型不满足需求。 """ def __init__(self, mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]): self.mean = mean self.std = std if not (isinstance(self.mean, list) and isinstance(self.std, list)): raise TypeError("NormalizeImage: input type is invalid.") from functools import reduce if reduce(lambda x, y: x * y, self.std) == 0: raise TypeError('NormalizeImage: std is invalid!') def __call__(self, im, im_info=None, label_info=None): """ Args: im (numnp.ndarraypy): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 """ mean = np.array(self.mean)[np.newaxis, np.newaxis, :] std = np.array(self.std)[np.newaxis, np.newaxis, :] im = normalize(im, mean, std) if label_info is None: return (im, im_info) else: return (im, im_info, label_info) class RandomDistort(DetTransform): """以一定的概率对图像进行随机像素内容变换,模型训练时的数据增强操作 1. 对变换的操作顺序进行随机化操作。 2. 按照1中的顺序以一定的概率在范围[-range, range]对图像进行随机像素内容变换。 Args: brightness_range (float): 明亮度因子的范围。默认为0.5。 brightness_prob (float): 随机调整明亮度的概率。默认为0.5。 contrast_range (float): 对比度因子的范围。默认为0.5。 contrast_prob (float): 随机调整对比度的概率。默认为0.5。 saturation_range (float): 饱和度因子的范围。默认为0.5。 saturation_prob (float): 随机调整饱和度的概率。默认为0.5。 hue_range (int): 色调因子的范围。默认为18。 hue_prob (float): 随机调整色调的概率。默认为0.5。 """ def __init__(self, brightness_range=0.5, brightness_prob=0.5, contrast_range=0.5, contrast_prob=0.5, saturation_range=0.5, saturation_prob=0.5, hue_range=18, hue_prob=0.5): self.brightness_range = brightness_range self.brightness_prob = brightness_prob self.contrast_range = contrast_range self.contrast_prob = contrast_prob self.saturation_range = saturation_range self.saturation_prob = saturation_prob self.hue_range = hue_range self.hue_prob = hue_prob def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 """ brightness_lower = 1 - self.brightness_range brightness_upper = 1 + self.brightness_range contrast_lower = 1 - self.contrast_range contrast_upper = 1 + self.contrast_range saturation_lower = 1 - self.saturation_range saturation_upper = 1 + self.saturation_range hue_lower = -self.hue_range hue_upper = self.hue_range ops = [brightness, contrast, saturation, hue] random.shuffle(ops) params_dict = { 'brightness': { 'brightness_lower': brightness_lower, 'brightness_upper': brightness_upper }, 'contrast': { 'contrast_lower': contrast_lower, 'contrast_upper': contrast_upper }, 'saturation': { 'saturation_lower': saturation_lower, 'saturation_upper': saturation_upper }, 'hue': { 'hue_lower': hue_lower, 'hue_upper': hue_upper } } prob_dict = { 'brightness': self.brightness_prob, 'contrast': self.contrast_prob, 'saturation': self.saturation_prob, 'hue': self.hue_prob } for id in range(4): params = params_dict[ops[id].__name__] prob = prob_dict[ops[id].__name__] params['im'] = im if np.random.uniform(0, 1) < prob: im = ops[id](**params) if label_info is None: return (im, im_info) else: return (im, im_info, label_info) class MixupImage(DetTransform): """对图像进行mixup操作,模型训练时的数据增强操作,目前仅YOLOv3模型支持该transform。 当label_info中不存在mixup字段时,直接返回,否则进行下述操作: 1. 从随机beta分布中抽取出随机因子factor。 2. - 当factor>=1.0时,去除label_info中的mixup字段,直接返回。 - 当factor<=0.0时,直接返回label_info中的mixup字段,并在label_info中去除该字段。 - 其余情况,执行下述操作: (1)原图像乘以factor,mixup图像乘以(1-factor),叠加2个结果。 (2)拼接原图像标注框和mixup图像标注框。 (3)拼接原图像标注框类别和mixup图像标注框类别。 (4)原图像标注框混合得分乘以factor,mixup图像标注框混合得分乘以(1-factor),叠加2个结果。 3. 更新im_info中的image_shape信息。 Args: alpha (float): 随机beta分布的下限。默认为1.5。 beta (float): 随机beta分布的上限。默认为1.5。 mixup_epoch (int): 在前mixup_epoch轮使用mixup增强操作;当该参数为-1时,该策略不会生效。 默认为-1。 Raises: ValueError: 数据长度不匹配。 """ def __init__(self, alpha=1.5, beta=1.5, mixup_epoch=-1): self.alpha = alpha self.beta = beta if self.alpha <= 0.0: raise ValueError("alpha shold be positive in MixupImage") if self.beta <= 0.0: raise ValueError("beta shold be positive in MixupImage") self.mixup_epoch = mixup_epoch def _mixup_img(self, img1, img2, factor): h = max(img1.shape[0], img2.shape[0]) w = max(img1.shape[1], img2.shape[1]) img = np.zeros((h, w, img1.shape[2]), 'float32') img[:img1.shape[0], :img1.shape[1], :] = \ img1.astype('float32') * factor img[:img2.shape[0], :img2.shape[1], :] += \ img2.astype('float32') * (1.0 - factor) return img.astype('float32') def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 其中,im_info更新字段为: - image_shape (np.ndarray): mixup后的图像高、宽二者组成的np.ndarray,形状为(2,)。 im_info删除的字段: - mixup (list): 与当前字段进行mixup的图像相关信息。 label_info更新字段为: - gt_bbox (np.ndarray): mixup后真实标注框坐标,形状为(n, 4), 其中n代表真实标注框的个数。 - gt_class (np.ndarray): mixup后每个真实标注框对应的类别序号,形状为(n, 1), 其中n代表真实标注框的个数。 - gt_score (np.ndarray): mixup后每个真实标注框对应的混合得分,形状为(n, 1), 其中n代表真实标注框的个数。 Raises: TypeError: 形参数据类型不满足需求。 """ if im_info is None: raise TypeError('Cannot do MixupImage! ' + 'Becasuse the im_info can not be None!') if 'mixup' not in im_info: if label_info is None: return (im, im_info) else: return (im, im_info, label_info) factor = np.random.beta(self.alpha, self.beta) factor = max(0.0, min(1.0, factor)) if im_info['epoch'] > self.mixup_epoch \ or factor >= 1.0: im_info.pop('mixup') if label_info is None: return (im, im_info) else: return (im, im_info, label_info) if factor <= 0.0: return im_info.pop('mixup') im = self._mixup_img(im, im_info['mixup'][0], factor) if label_info is None: raise TypeError('Cannot do MixupImage! ' + 'Becasuse the label_info can not be None!') if 'gt_bbox' not in label_info or \ 'gt_class' not in label_info or \ 'gt_score' not in label_info: raise TypeError('Cannot do MixupImage! ' + \ 'Becasuse gt_bbox/gt_class/gt_score is not in label_info!') gt_bbox1 = label_info['gt_bbox'] gt_bbox2 = im_info['mixup'][2]['gt_bbox'] gt_class1 = label_info['gt_class'] gt_class2 = im_info['mixup'][2]['gt_class'] gt_score1 = label_info['gt_score'] gt_score2 = im_info['mixup'][2]['gt_score'] if 'gt_poly' in label_info: gt_poly1 = label_info['gt_poly'] gt_poly2 = im_info['mixup'][2]['gt_poly'] is_crowd1 = label_info['is_crowd'] is_crowd2 = im_info['mixup'][2]['is_crowd'] if 0 not in gt_class1 and 0 not in gt_class2: gt_bbox = np.concatenate((gt_bbox1, gt_bbox2), axis=0) gt_class = np.concatenate((gt_class1, gt_class2), axis=0) gt_score = np.concatenate( (gt_score1 * factor, gt_score2 * (1. - factor)), axis=0) if 'gt_poly' in label_info: label_info['gt_poly'] = gt_poly1 + gt_poly2 is_crowd = np.concatenate((is_crowd1, is_crowd2), axis=0) elif 0 in gt_class1: gt_bbox = gt_bbox2 gt_class = gt_class2 gt_score = gt_score2 * (1. - factor) if 'gt_poly' in label_info: label_info['gt_poly'] = gt_poly2 is_crowd = is_crowd2 else: gt_bbox = gt_bbox1 gt_class = gt_class1 gt_score = gt_score1 * factor if 'gt_poly' in label_info: label_info['gt_poly'] = gt_poly1 is_crowd = is_crowd1 label_info['gt_bbox'] = gt_bbox label_info['gt_score'] = gt_score label_info['gt_class'] = gt_class label_info['is_crowd'] = is_crowd im_info['image_shape'] = np.array([im.shape[0], im.shape[1]]).astype('int32') im_info.pop('mixup') if label_info is None: return (im, im_info) else: return (im, im_info, label_info) class RandomExpand(DetTransform): """随机扩张图像,模型训练时的数据增强操作。 1. 随机选取扩张比例(扩张比例大于1时才进行扩张)。 2. 计算扩张后图像大小。 3. 初始化像素值为输入填充值的图像,并将原图像随机粘贴于该图像上。 4. 根据原图像粘贴位置换算出扩张后真实标注框的位置坐标。 5. 根据原图像粘贴位置换算出扩张后真实分割区域的位置坐标。 Args: ratio (float): 图像扩张的最大比例。默认为4.0。 prob (float): 随机扩张的概率。默认为0.5。 fill_value (list): 扩张图像的初始填充值(0-255)。默认为[123.675, 116.28, 103.53]。 """ def __init__(self, ratio=4., prob=0.5, fill_value=[123.675, 116.28, 103.53]): super(RandomExpand, self).__init__() assert ratio > 1.01, "expand ratio must be larger than 1.01" self.ratio = ratio self.prob = prob assert isinstance(fill_value, Sequence), \ "fill value must be sequence" if not isinstance(fill_value, tuple): fill_value = tuple(fill_value) self.fill_value = fill_value def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 其中,im_info更新字段为: - image_shape (np.ndarray): 扩张后的图像高、宽二者组成的np.ndarray,形状为(2,)。 label_info更新字段为: - gt_bbox (np.ndarray): 随机扩张后真实标注框坐标,形状为(n, 4), 其中n代表真实标注框的个数。 - gt_class (np.ndarray): 随机扩张后每个真实标注框对应的类别序号,形状为(n, 1), 其中n代表真实标注框的个数。 Raises: TypeError: 形参数据类型不满足需求。 """ if im_info is None or label_info is None: raise TypeError( 'Cannot do RandomExpand! ' + 'Becasuse the im_info and label_info can not be None!') if 'gt_bbox' not in label_info or \ 'gt_class' not in label_info: raise TypeError('Cannot do RandomExpand! ' + \ 'Becasuse gt_bbox/gt_class is not in label_info!') if np.random.uniform(0., 1.) < self.prob: return (im, im_info, label_info) if 'gt_class' in label_info and 0 in label_info['gt_class']: return (im, im_info, label_info) image_shape = im_info['image_shape'] height = int(image_shape[0]) width = int(image_shape[1]) expand_ratio = np.random.uniform(1., self.ratio) h = int(height * expand_ratio) w = int(width * expand_ratio) if not h > height or not w > width: return (im, im_info, label_info) y = np.random.randint(0, h - height) x = np.random.randint(0, w - width) canvas = np.ones((h, w, 3), dtype=np.float32) canvas *= np.array(self.fill_value, dtype=np.float32) canvas[y:y + height, x:x + width, :] = im im_info['image_shape'] = np.array([h, w]).astype('int32') if 'gt_bbox' in label_info and len(label_info['gt_bbox']) > 0: label_info['gt_bbox'] += np.array([x, y] * 2, dtype=np.float32) if 'gt_poly' in label_info and len(label_info['gt_poly']) > 0: label_info['gt_poly'] = expand_segms(label_info['gt_poly'], x, y, height, width, expand_ratio) return (canvas, im_info, label_info) class RandomCrop(DetTransform): """随机裁剪图像。 1. 若allow_no_crop为True,则在thresholds加入’no_crop’。 2. 随机打乱thresholds。 3. 遍历thresholds中各元素: (1) 如果当前thresh为’no_crop’,则返回原始图像和标注信息。 (2) 随机取出aspect_ratio和scaling中的值并由此计算出候选裁剪区域的高、宽、起始点。 (3) 计算真实标注框与候选裁剪区域IoU,若全部真实标注框的IoU都小于thresh,则继续第3步。 (4) 如果cover_all_box为True且存在真实标注框的IoU小于thresh,则继续第3步。 (5) 筛选出位于候选裁剪区域内的真实标注框,若有效框的个数为0,则继续第3步,否则进行第4步。 4. 换算有效真值标注框相对候选裁剪区域的位置坐标。 5. 换算有效分割区域相对候选裁剪区域的位置坐标。 Args: aspect_ratio (list): 裁剪后短边缩放比例的取值范围,以[min, max]形式表示。默认值为[.5, 2.]。 thresholds (list): 判断裁剪候选区域是否有效所需的IoU阈值取值列表。默认值为[.0, .1, .3, .5, .7, .9]。 scaling (list): 裁剪面积相对原面积的取值范围,以[min, max]形式表示。默认值为[.3, 1.]。 num_attempts (int): 在放弃寻找有效裁剪区域前尝试的次数。默认值为50。 allow_no_crop (bool): 是否允许未进行裁剪。默认值为True。 cover_all_box (bool): 是否要求所有的真实标注框都必须在裁剪区域内。默认值为False。 """ def __init__(self, aspect_ratio=[.5, 2.], thresholds=[.0, .1, .3, .5, .7, .9], scaling=[.3, 1.], num_attempts=50, allow_no_crop=True, cover_all_box=False): self.aspect_ratio = aspect_ratio self.thresholds = thresholds self.scaling = scaling self.num_attempts = num_attempts self.allow_no_crop = allow_no_crop self.cover_all_box = cover_all_box def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当label_info为空时,返回的tuple为(im, im_info),分别对应图像np.ndarray数据、存储与图像相关信息的字典; 当label_info不为空时,返回的tuple为(im, im_info, label_info),分别对应图像np.ndarray数据、 存储与标注框相关信息的字典。 其中,im_info更新字段为: - image_shape (np.ndarray): 扩裁剪的图像高、宽二者组成的np.ndarray,形状为(2,)。 label_info更新字段为: - gt_bbox (np.ndarray): 随机裁剪后真实标注框坐标,形状为(n, 4), 其中n代表真实标注框的个数。 - gt_class (np.ndarray): 随机裁剪后每个真实标注框对应的类别序号,形状为(n, 1), 其中n代表真实标注框的个数。 - gt_score (np.ndarray): 随机裁剪后每个真实标注框对应的混合得分,形状为(n, 1), 其中n代表真实标注框的个数。 Raises: TypeError: 形参数据类型不满足需求。 """ if im_info is None or label_info is None: raise TypeError( 'Cannot do RandomCrop! ' + 'Becasuse the im_info and label_info can not be None!') if 'gt_bbox' not in label_info or \ 'gt_class' not in label_info: raise TypeError('Cannot do RandomCrop! ' + \ 'Becasuse gt_bbox/gt_class is not in label_info!') if len(label_info['gt_bbox']) == 0: return (im, im_info, label_info) if 'gt_class' in label_info and 0 in label_info['gt_class']: return (im, im_info, label_info) image_shape = im_info['image_shape'] w = image_shape[1] h = image_shape[0] gt_bbox = label_info['gt_bbox'] thresholds = list(self.thresholds) if self.allow_no_crop: thresholds.append('no_crop') np.random.shuffle(thresholds) for thresh in thresholds: if thresh == 'no_crop': return (im, im_info, label_info) found = False for i in range(self.num_attempts): scale = np.random.uniform(*self.scaling) min_ar, max_ar = self.aspect_ratio aspect_ratio = np.random.uniform( max(min_ar, scale**2), min(max_ar, scale**-2)) crop_h = int(h * scale / np.sqrt(aspect_ratio)) crop_w = int(w * scale * np.sqrt(aspect_ratio)) crop_y = np.random.randint(0, h - crop_h) crop_x = np.random.randint(0, w - crop_w) crop_box = [crop_x, crop_y, crop_x + crop_w, crop_y + crop_h] iou = iou_matrix( gt_bbox, np.array( [crop_box], dtype=np.float32)) if iou.max() < thresh: continue if self.cover_all_box and iou.min() < thresh: continue cropped_box, valid_ids = crop_box_with_center_constraint( gt_bbox, np.array( crop_box, dtype=np.float32)) if valid_ids.size > 0: found = True break if found: if 'gt_poly' in label_info and len(label_info['gt_poly']) > 0: crop_polys = crop_segms( label_info['gt_poly'], valid_ids, np.array( crop_box, dtype=np.int64), h, w) if [] in crop_polys: delete_id = list() valid_polys = list() for id, crop_poly in enumerate(crop_polys): if crop_poly == []: delete_id.append(id) else: valid_polys.append(crop_poly) valid_ids = np.delete(valid_ids, delete_id) if len(valid_polys) == 0: return (im, im_info, label_info) label_info['gt_poly'] = valid_polys else: label_info['gt_poly'] = crop_polys im = crop_image(im, crop_box) label_info['gt_bbox'] = np.take(cropped_box, valid_ids, axis=0) label_info['gt_class'] = np.take( label_info['gt_class'], valid_ids, axis=0) im_info['image_shape'] = np.array( [crop_box[3] - crop_box[1], crop_box[2] - crop_box[0]]).astype('int32') if 'gt_score' in label_info: label_info['gt_score'] = np.take( label_info['gt_score'], valid_ids, axis=0) if 'is_crowd' in label_info: label_info['is_crowd'] = np.take( label_info['is_crowd'], valid_ids, axis=0) return (im, im_info, label_info) return (im, im_info, label_info) class ArrangeFasterRCNN(DetTransform): """获取FasterRCNN模型训练/验证/预测所需信息。 Args: mode (str): 指定数据用于何种用途,取值范围为['train', 'eval', 'test', 'quant']。 Raises: ValueError: mode的取值不在['train', 'eval', 'test', 'quant']之内。 """ def __init__(self, mode=None): if mode not in ['train', 'eval', 'test', 'quant']: raise ValueError( "mode must be in ['train', 'eval', 'test', 'quant']!") self.mode = mode def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当mode为'train'时,返回(im, im_resize_info, gt_bbox, gt_class, is_crowd),分别对应 图像np.ndarray数据、图像相当对于原图的resize信息、真实标注框、真实标注框对应的类别、真实标注框内是否是一组对象; 当mode为'eval'时,返回(im, im_resize_info, im_id, im_shape, gt_bbox, gt_class, is_difficult), 分别对应图像np.ndarray数据、图像相当对于原图的resize信息、图像id、图像大小信息、真实标注框、真实标注框对应的类别、 真实标注框是否为难识别对象;当mode为'test'或'quant'时,返回(im, im_resize_info, im_shape),分别对应图像np.ndarray数据、 图像相当对于原图的resize信息、图像大小信息。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 数据长度不匹配。 """ im = permute(im, False) if self.mode == 'train': if im_info is None or label_info is None: raise TypeError( 'Cannot do ArrangeFasterRCNN! ' + 'Becasuse the im_info and label_info can not be None!') if len(label_info['gt_bbox']) != len(label_info['gt_class']): raise ValueError("gt num mismatch: bbox and class.") im_resize_info = im_info['im_resize_info'] gt_bbox = label_info['gt_bbox'] gt_class = label_info['gt_class'] is_crowd = label_info['is_crowd'] outputs = (im, im_resize_info, gt_bbox, gt_class, is_crowd) elif self.mode == 'eval': if im_info is None or label_info is None: raise TypeError( 'Cannot do ArrangeFasterRCNN! ' + 'Becasuse the im_info and label_info can not be None!') im_resize_info = im_info['im_resize_info'] im_id = im_info['im_id'] im_shape = np.array( (im_info['image_shape'][0], im_info['image_shape'][1], 1), dtype=np.float32) gt_bbox = label_info['gt_bbox'] gt_class = label_info['gt_class'] is_difficult = label_info['difficult'] outputs = (im, im_resize_info, im_id, im_shape, gt_bbox, gt_class, is_difficult) else: if im_info is None: raise TypeError('Cannot do ArrangeFasterRCNN! ' + 'Becasuse the im_info can not be None!') im_resize_info = im_info['im_resize_info'] im_shape = np.array( (im_info['image_shape'][0], im_info['image_shape'][1], 1), dtype=np.float32) outputs = (im, im_resize_info, im_shape) return outputs class ArrangeMaskRCNN(DetTransform): """获取MaskRCNN模型训练/验证/预测所需信息。 Args: mode (str): 指定数据用于何种用途,取值范围为['train', 'eval', 'test', 'quant']。 Raises: ValueError: mode的取值不在['train', 'eval', 'test', 'quant']之内。 """ def __init__(self, mode=None): if mode not in ['train', 'eval', 'test', 'quant']: raise ValueError( "mode must be in ['train', 'eval', 'test', 'quant']!") self.mode = mode def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当mode为'train'时,返回(im, im_resize_info, gt_bbox, gt_class, is_crowd, gt_masks),分别对应 图像np.ndarray数据、图像相当对于原图的resize信息、真实标注框、真实标注框对应的类别、真实标注框内是否是一组对象、 真实分割区域;当mode为'eval'时,返回(im, im_resize_info, im_id, im_shape),分别对应图像np.ndarray数据、 图像相当对于原图的resize信息、图像id、图像大小信息;当mode为'test'或'quant'时,返回(im, im_resize_info, im_shape), 分别对应图像np.ndarray数据、图像相当对于原图的resize信息、图像大小信息。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 数据长度不匹配。 """ im = permute(im, False) if self.mode == 'train': if im_info is None or label_info is None: raise TypeError( 'Cannot do ArrangeTrainMaskRCNN! ' + 'Becasuse the im_info and label_info can not be None!') if len(label_info['gt_bbox']) != len(label_info['gt_class']): raise ValueError("gt num mismatch: bbox and class.") im_resize_info = im_info['im_resize_info'] gt_bbox = label_info['gt_bbox'] gt_class = label_info['gt_class'] is_crowd = label_info['is_crowd'] assert 'gt_poly' in label_info segms = label_info['gt_poly'] if len(segms) != 0: assert len(segms) == is_crowd.shape[0] gt_masks = [] valid = True for i in range(len(segms)): segm = segms[i] gt_segm = [] if is_crowd[i]: gt_segm.append([[0, 0]]) else: for poly in segm: if len(poly) == 0: valid = False break gt_segm.append(np.array(poly).reshape(-1, 2)) if (not valid) or len(gt_segm) == 0: break gt_masks.append(gt_segm) outputs = (im, im_resize_info, gt_bbox, gt_class, is_crowd, gt_masks) else: if im_info is None: raise TypeError('Cannot do ArrangeMaskRCNN! ' + 'Becasuse the im_info can not be None!') im_resize_info = im_info['im_resize_info'] im_shape = np.array( (im_info['image_shape'][0], im_info['image_shape'][1], 1), dtype=np.float32) if self.mode == 'eval': im_id = im_info['im_id'] outputs = (im, im_resize_info, im_id, im_shape) else: outputs = (im, im_resize_info, im_shape) return outputs class ArrangeYOLOv3(DetTransform): """获取YOLOv3模型训练/验证/预测所需信息。 Args: mode (str): 指定数据用于何种用途,取值范围为['train', 'eval', 'test', 'quant']。 Raises: ValueError: mode的取值不在['train', 'eval', 'test', 'quant']之内。 """ def __init__(self, mode=None): if mode not in ['train', 'eval', 'test', 'quant']: raise ValueError( "mode must be in ['train', 'eval', 'test', 'quant']!") self.mode = mode def __call__(self, im, im_info=None, label_info=None): """ Args: im (np.ndarray): 图像np.ndarray数据。 im_info (dict, 可选): 存储与图像相关的信息。 label_info (dict, 可选): 存储与标注框相关的信息。 Returns: tuple: 当mode为'train'时,返回(im, gt_bbox, gt_class, gt_score, im_shape),分别对应 图像np.ndarray数据、真实标注框、真实标注框对应的类别、真实标注框混合得分、图像大小信息; 当mode为'eval'时,返回(im, im_shape, im_id, gt_bbox, gt_class, difficult), 分别对应图像np.ndarray数据、图像大小信息、图像id、真实标注框、真实标注框对应的类别、 真实标注框是否为难识别对象;当mode为'test'或'quant'时,返回(im, im_shape), 分别对应图像np.ndarray数据、图像大小信息。 Raises: TypeError: 形参数据类型不满足需求。 ValueError: 数据长度不匹配。 """ im = permute(im, False) if self.mode == 'train': if im_info is None or label_info is None: raise TypeError( 'Cannot do ArrangeYolov3! ' + 'Becasuse the im_info and label_info can not be None!') im_shape = im_info['image_shape'] if len(label_info['gt_bbox']) != len(label_info['gt_class']): raise ValueError("gt num mismatch: bbox and class.") if len(label_info['gt_bbox']) != len(label_info['gt_score']): raise ValueError("gt num mismatch: bbox and score.") gt_bbox = np.zeros((50, 4), dtype=im.dtype) gt_class = np.zeros((50, ), dtype=np.int32) gt_score = np.zeros((50, ), dtype=im.dtype) gt_num = min(50, len(label_info['gt_bbox'])) if gt_num > 0: label_info['gt_class'][:gt_num, 0] = label_info[ 'gt_class'][:gt_num, 0] - 1 if -1 not in label_info['gt_class']: gt_bbox[:gt_num, :] = label_info['gt_bbox'][:gt_num, :] gt_class[:gt_num] = label_info['gt_class'][:gt_num, 0] gt_score[:gt_num] = label_info['gt_score'][:gt_num, 0] # parse [x1, y1, x2, y2] to [x, y, w, h] gt_bbox[:, 2:4] = gt_bbox[:, 2:4] - gt_bbox[:, :2] gt_bbox[:, :2] = gt_bbox[:, :2] + gt_bbox[:, 2:4] / 2. outputs = (im, gt_bbox, gt_class, gt_score, im_shape) elif self.mode == 'eval': if im_info is None or label_info is None: raise TypeError( 'Cannot do ArrangeYolov3! ' + 'Becasuse the im_info and label_info can not be None!') im_shape = im_info['image_shape'] if len(label_info['gt_bbox']) != len(label_info['gt_class']): raise ValueError("gt num mismatch: bbox and class.") im_id = im_info['im_id'] gt_bbox = np.zeros((50, 4), dtype=im.dtype) gt_class = np.zeros((50, ), dtype=np.int32) difficult = np.zeros((50, ), dtype=np.int32) gt_num = min(50, len(label_info['gt_bbox'])) if gt_num > 0: label_info['gt_class'][:gt_num, 0] = label_info[ 'gt_class'][:gt_num, 0] - 1 gt_bbox[:gt_num, :] = label_info['gt_bbox'][:gt_num, :] gt_class[:gt_num] = label_info['gt_class'][:gt_num, 0] difficult[:gt_num] = label_info['difficult'][:gt_num, 0] outputs = (im, im_shape, im_id, gt_bbox, gt_class, difficult) else: if im_info is None: raise TypeError('Cannot do ArrangeYolov3! ' + 'Becasuse the im_info can not be None!') im_shape = im_info['image_shape'] outputs = (im, im_shape) return outputs class ComposedRCNNTransforms(Compose): """ RCNN模型(faster-rcnn/mask-rcnn)图像处理流程,具体如下, 训练阶段: 1. 随机以0.5的概率将图像水平翻转 2. 图像归一化 3. 图像按比例Resize,scale计算方式如下 scale = min_max_size[0] / short_size_of_image if max_size_of_image * scale > min_max_size[1]: scale = min_max_size[1] / max_size_of_image 4. 将3步骤的长宽进行padding,使得长宽为32的倍数 验证阶段: 1. 图像归一化 2. 图像按比例Resize,scale计算方式同上训练阶段 3. 将2步骤的长宽进行padding,使得长宽为32的倍数 Args: mode(str): 图像处理流程所处阶段,训练/验证/预测,分别对应'train', 'eval', 'test' min_max_size(list): 图像在缩放时,最小边和最大边的约束条件 mean(list): 图像均值 std(list): 图像方差 """ def __init__(self, mode, min_max_size=[800, 1333], mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]): if mode == 'train': # 训练时的transforms,包含数据增强 transforms = [ RandomHorizontalFlip(prob=0.5), Normalize( mean=mean, std=std), ResizeByShort( short_size=min_max_size[0], max_size=min_max_size[1]), Padding(coarsest_stride=32) ] else: # 验证/预测时的transforms transforms = [ Normalize( mean=mean, std=std), ResizeByShort( short_size=min_max_size[0], max_size=min_max_size[1]), Padding(coarsest_stride=32) ] super(ComposedRCNNTransforms, self).__init__(transforms) class ComposedYOLOv3Transforms(Compose): """YOLOv3模型的图像预处理流程,具体如下, 训练阶段: 1. 在前mixup_epoch轮迭代中,使用MixupImage策略,见https://paddlex.readthedocs.io/zh_CN/latest/apis/transforms/det_transforms.html#mixupimage 2. 对图像进行随机扰动,包括亮度,对比度,饱和度和色调 3. 随机扩充图像,见https://paddlex.readthedocs.io/zh_CN/latest/apis/transforms/det_transforms.html#randomexpand 4. 随机裁剪图像 5. 将4步骤的输出图像Resize成shape参数的大小 6. 随机0.5的概率水平翻转图像 7. 图像归一化 验证/预测阶段: 1. 将图像Resize成shape参数大小 2. 图像归一化 Args: mode(str): 图像处理流程所处阶段,训练/验证/预测,分别对应'train', 'eval', 'test' shape(list): 输入模型中图像的大小,输入模型的图像会被Resize成此大小 mixup_epoch(int): 模型训练过程中,前mixup_epoch会使用mixup策略 mean(list): 图像均值 std(list): 图像方差 """ def __init__(self, mode, shape=[608, 608], mixup_epoch=250, mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]): width = shape if isinstance(shape, list): if shape[0] != shape[1]: raise Exception( "In YOLOv3 model, width and height should be equal") width = shape[0] if width % 32 != 0: raise Exception( "In YOLOv3 model, width and height should be multiple of 32, e.g 224、256、320...." ) if mode == 'train': # 训练时的transforms,包含数据增强 transforms = [ MixupImage(mixup_epoch=mixup_epoch), RandomDistort(), RandomExpand(), RandomCrop(), Resize( target_size=width, interp='RANDOM'), RandomHorizontalFlip(), Normalize( mean=mean, std=std) ] else: # 验证/预测时的transforms transforms = [ Resize( target_size=width, interp='CUBIC'), Normalize( mean=mean, std=std) ] super(ComposedYOLOv3Transforms, self).__init__(transforms)