def test_tf(img):
img = tfs.Resize(256)(img)
img, _ = tfs.CenterCrop(224)(img)
normalize = tfs.Compose([
tfs.ToTensor(),
tfs.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])
img = normalize(img)
return img
def img_transforms(img, label, crop_size):
img, label = random_crop(img, label, crop_size)
img_tfs = tfs.Compose([
tfs.ToTensor(),
tfs.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])
img = img_tfs(img)
label = image2label(label)
label = torch.from_numpy(label)
return img, label
def pytorch_normalize(img):
"""Normalize input image in pytorch way.
Use pytorch way to do normalize, taking from a numpy.ndarray.
Args:
img(~numpy.ndarray): Image to be normalized, which is a numpy.ndarray.
Returns:
An normalized torch.Tensor, CHW order and range from :math:`[0, 1]`.
"""
img = torch.from_numpy(img) / 255.
normalize = tfs.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
img = normalize(img)
return img.numpy()
from config import opt
from mxtorch import meter
from mxtorch import transforms as tfs
from mxtorch.trainer import *
from mxtorch.vision import model_zoo
from torch import nn
from torch.autograd import Variable
from torch.utils.data import DataLoader
from torchvision.datasets import ImageFolder
from tqdm import tqdm
train_tf = tfs.Compose([
tfs.RandomResizedCrop(224),
tfs.RandomHorizontalFlip(),
tfs.ToTensor(),
tfs.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])
def test_tf(img):
img = tfs.Resize(256)(img)
img, _ = tfs.CenterCrop(224)(img)
normalize = tfs.Compose([
tfs.ToTensor(),
tfs.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])
img = normalize(img)
return img
def get_train_data():