def __init__(self, mode='train'):
"""
初始化函数
"""
assert mode in ['train', 'test',
'valid'], 'mode is one of train, test, valid.'
self.data = []
with open('signs/{}.txt'.format(mode)) as f:
for line in f.readlines():
info = line.strip().split('\t')
if len(info) > 0:
self.data.append([info[0].strip(), info[1].strip()])
if mode == 'train':
self.transforms = T.Compose([
T.RandomResizedCrop(IMAGE_SIZE), # 随机裁剪大小
T.RandomHorizontalFlip(0.5), # 随机水平翻转
T.ToTensor(), # 数据的格式转换和标准化 HWC => CHW
T.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]) # 图像归一化
])
else:
self.transforms = T.Compose([
T.Resize(256), # 图像大小修改
T.RandomCrop(IMAGE_SIZE), # 随机裁剪
T.ToTensor(), # 数据的格式转换和标准化 HWC => CHW
T.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]) # 图像归一化
])
def get_dataloader(self, num_workers):
dataset = paddle.vision.datasets.MNIST(
mode='test',
transform=transforms.Compose([
transforms.CenterCrop(20),
transforms.RandomResizedCrop(14),
transforms.Normalize(),
transforms.ToTensor()
]))
loader = paddle.io.DataLoader(dataset,
batch_size=32,
num_workers=num_workers,
shuffle=True)
return loader
from paddle.metric import Accuracy
from paddle.vision.models import mobilenet_v2
import warnings
warnings.filterwarnings("ignore")
from dataset import MyDataset
import paddle.vision.transforms as T
transform = T.Compose([
T.RandomResizedCrop([448, 448]),
T.RandomHorizontalFlip(),
T.RandomRotation(90),
T.ToTensor(),
T.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]),
])
train_dataset = MyDataset(txt=r'C:\Users\11982\Desktop\paddle\train.txt', transform=transform)
train_loader = paddle.io.DataLoader(train_dataset, places=paddle.CPUPlace(), batch_size=8, shuffle=True, num_workers=8)
# build model
model = mobilenet_v2(pretrained=True,scale=1.0, num_classes=2, with_pool=True)
# 自定义Callback 记录训练过程中的loss信息
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# author:gentelyang time:2021-06-10
import os
import paddle
import paddle.distributed as dist
from paddle.io import DataLoader
from paddle.vision import transforms
normalize = transforms.Normalize(mean=[123.675, 116.28, 103.53],
std=[58.395, 57.120, 57.375])
transform = transforms.Compose([
transforms.RandomResizedCrop(224),
transforms.RandomHorizontalFlip(),
transforms.Transpose(), normalize
])
def train():
# 设置支持多卡训练
dist.init_parallel_env()
train_dataset = paddle.vision.datasets.Cifar10(mode='train',
transform=transform)
batch_sampler = paddle.io.DistributedBatchSampler(train_dataset,
batch_size=32,
shuffle=True)
train_loader = DataLoader(dataset=train_dataset,
batch_sampler=batch_sampler)
model = paddle.vision.mobilenet_v2(num_classes=10)
# 设置支持多卡训练
model = paddle.DataParallel(model)
# 设置优化方法
p.requires_grad = val
# augmentation utils
normalize = T.Normalize(mean=[0.485 * 255, 0.456 * 255, 0.406 * 255],
std=[0.229 * 255, 0.224 * 255, 0.225 * 255],
data_format='CHW'),
augmentation = [
byol.transforms.RandomApply([
T.ColorJitter(0.8, 0.8, 0.8, 0.2),
], p=0.3),
byol.transforms.RandomGrayscale(p=0.2),
byol.transforms.RandomApply([byol.transforms.GaussianBlur((1.0, 2.0))],
p=0.2),
T.RandomResizedCrop(224, scale=(0.2, 1.)),
# T.RandomResizedCrop((image_size, image_size)),
normalize
]
# MLP class for projector and predictor
class MLP(nn.Layer):
def __init__(self, dim, projection_size, hidden_size=4096):
super(MLP, self).__init__()
self.net = nn.Sequential(nn.Linear(dim, hidden_size),
nn.BatchNorm1D(hidden_size), nn.ReLU(),
nn.Linear(hidden_size, projection_size))
def forward(self, x):
