matlabud.save,YOLOV5训练与测试时数据加载dataset.py代
码注释与解析
import glob
import math
import os
import random
import shutil
import time
from pathlib import Path
from threading import Thread
import cv2
import numpy as np
import torch
from PIL import Image, ExifTags
from torch.utils.data import Dataset
resized
from tqdm import tqdm
from utils.utils import xyxy2xywh, xywh2xyxy
help_url = 'github/ultralytics/yolov5/wiki/Train-Custom-Data'
img_formats = ['.bmp', '.jpg', '.jpeg', '.png', '.tif', '.dng']
vid_formats = ['.mov', '.avi', '.mp4', '.mpg', '.mpeg', '.m4v', '.wmv', '.mkv']
# Get orientation exif tag
for orientation in ExifTags.TAGS.keys():
if ExifTags.TAGS[orientation] == 'Orientation':
break
# 此函数根据图⽚的信息获取图⽚的宽、⾼信息
def exif_size(img):
# Returns exif-corrected PIL size
s = img.size # (width, height)
try:
rotation = dict(img._getexif().items())[orientation]
if rotation == 6: # rotation 270
s = (s[1], s[0])
elif rotation == 8: # rotation 90
s = (s[1], s[0])
except:
pass
return s
# 根据LoadImagesAndLabels创建dataloader
def create_dataloader(path, imgsz, batch_size, stride, opt, hyp=None, augment=False, cache=False, pad=0.0, rect=False): """
参数解析：
path：包含图⽚路径的txt⽂件或者包含图⽚的⽂件夹路径
imgsz：⽹络输⼊图⽚⼤⼩
batch_size: 批次⼤⼩
stride：⽹络下采样最⼤总步长
opt：调⽤train.py时传⼊的参数，这⾥主要⽤到opt.single_cls，是否是单类数据集
hyp：⽹络训练时的⼀些超参数，包括学习率等，这⾥主要⽤到⾥⾯⼀些关于数据增强(旋转、平移等)的系数
augment：是否进⾏数据增强
cache：是否提前缓存图⽚到内存，以便加快训练速度
pad：设置矩形训练的shape时进⾏的填充
rect：是否进⾏矩形训练
"""
dataset = LoadImagesAndLabels(path, imgsz, batch_size,
augment=augment, # augment images
hyp=hyp, # augmentation hyperparameters
rect=rect, # rectangular training
cache_images=cache,
single_cls=opt.single_cls,
stride=int(stride),
pad=pad)
batch_size = min(batch_size, len(dataset))
nw = min([os.cpu_count(), batch_size if batch_size > 1 else 0, 8]) # number of workers
dataloader = torch.utils.data.DataLoader(dataset,
batch_size=batch_size,
num_workers=nw,
pin_memory=True,
collate_llate_fn)
return dataloader, dataset
class LoadImagesAndLabels(Dataset): # for training/testing
def __init__(self, path, img_size=640, batch_size=16, augment=False, hyp=None, rect=False, image_weights=False, cache_images=False, single_cls=False, stride=32, pad=0.0):
try:
f = []
for p in path if isinstance(path, list) else [path]:
# 获取数据集路径path，包含图⽚路径的txt⽂件或者包含图⽚的⽂件夹路径
# 使⽤pathlib.Path⽣成与操作系统⽆关的路径，因为不同操作系统路径的‘/’会有所不同
p = str(Path(p)) # os-agnostic
# 获取数据集路径的上级⽗⽬录，os.sep为路径⾥的破折号(不同系统路径破折号不同，os.sep根据系统⾃适应)
parent = str(Path(p).parent) + os.sep
# 如果路径path为包含图⽚路径的txt⽂件
if os.path.isfile(p): # file
with open(p, 'r') as t:
# 获取图⽚路径，更换相对路径
t = t.read().splitlines()
f += [x.replace('./', parent) if x.startswith('./') else x for x in t] # local to global path
# 如果路径path为包含图⽚的⽂件夹路径
elif os.path.isdir(p): # folder
f += glob.iglob(p + os.sep + '*.*')
else:
raise Exception('%s does not exist' % p)
path = p # *.npy dir
# 破折号替换为os.sep，os.path.splitext(x)将⽂件名与扩展名分开并返回⼀个列表
self.img_files = [x.replace('/', os.sep) for x in f if os.path.splitext(x)[-1].lower() in img_formats]
except Exception as e:
raise Exception('Error loading data from %s: %s\nSee %s' % (path, e, help_url))
# 数据集的数量
n = len(self.img_files)
assert n > 0, 'No images found in %s. See %s' % (path, help_url)
# 获取batch的索引
bi = np.floor(np.arange(n) / batch_size).astype(np.int) # batch index
# ⼀个轮次batch的数量
nb = bi[-1] + 1 # number of batches
self.n = n # number of images
self.batch = bi # batch index of image
self.img_size = img_size # 输⼊图⽚分辨率⼤⼩
self.augment = augment # 数据增强
self.hyp = hyp # 超参数
self.image_weights = image_weights # 图⽚采样
< = False if image_weights else rect # 矩形训练
self.stride = stride # 模型下采样的总步长
# 获取数据集的标签
self.label_files = [x.replace('images', 'labels').replace(os.path.splitext(x)[-1], '.txt')
for x in self.img_files]
# 保存图⽚shape的路径
sp = place('.txt', '') + '.shapes' # shapefile path
try:
# 如果存在该路径，则读取
with open(sp, 'r') as f: # read existing shapefile
s = [x.split() for x ad().splitlines()]
assert len(s) == n, 'Shapefile out of sync'
except:
# 如果不存在，则读取图⽚shape再保存
s = [exif_size(Image.open(f)) for f in tqdm(self.img_files, desc='Reading image shapes')] np.savetxt(sp, s, fmt='%g') # overwrites existing (if any)
self.shapes = np.array(s, dtype=np.float64)
# 矩形训练
:
# Sort by aspect ratio
s = self.shapes # wh
ar = s[:, 1] / s[:, 0] # h/w
# 获取根据ar从⼩到⼤排序的索引
irect = ar.argsort()
# 根据索引排序数据集与标签路径、shape、h/w
self.img_files = [self.img_files[i] for i in irect]
self.label_files = [self.label_files[i] for i in irect]
self.shapes = s[irect] # wh
ar = ar[irect]
# Set training image shapes
# 初始化shapes，nb为⼀轮批次batch的数量
shapes = [[1, 1]] * nb
for i in range(nb):
ari = ar[bi == i]
mini, maxi = ari.min(), ari.max()
# 如果⼀个batch中最⼤的h/w⼩于1，则此batch的shape为(img_size*maxi, img_size)
if maxi < 1:
shapes[i] = [maxi, 1]
# 如果⼀个batch中最⼩的h/w⼤于1，则此batch的shape为(img_size, img_size/mini)
elif mini > 1:
shapes[i] = [1, 1 / mini]
self.batch_shapes = np.ceil(np.array(shapes) * img_size / stride + pad).astype(np.int) * stride # Cache labels
# 初始化图⽚与标签，为缓存图⽚、标签做准备
self.imgs = [None] * n
self.labels = [np.zeros((0, 5), dtype=np.float32)] * n
# 设置是否创建数据⼦集、提取⽬标检测框做再次分类，labels是否已加载
create_datasubset, extract_bounding_boxes, labels_loaded = False, False, False
# 漏掉的标签数量，到的标签数量，空的标签数量，数据⼦集的数量，相同的标签数量
nm, nf, ne, ns, nd = 0, 0, 0, 0, 0 # number missing, found, empty, datasubset, duplicate
# 保存labels的numpy⽂件路径
np_labels_path = str(Path(self.label_files[0]).parent) + '.npy' # saved labels in *.npy file
# 如果存在labels.npy，则直接加载，并设置labels_loaded=True
if os.path.isfile(np_labels_path):
s = np_labels_path # print string
x = np.load(np_labels_path, allow_pickle=True)
if len(x) == n:
self.labels = x
labels_loaded = True

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。