def Generate_transform_Dict(origin_width=256, width=227, ratio=0.16, RAE='None'): std_value = 1.0 / 255.0 normalize = transforms.Normalize( mean=[104 / 255.0, 117 / 255.0, 128 / 255.0], std=[1.0 / 255, 1.0 / 255, 1.0 / 255]) transform_dict = {} if RAE == 'None': print("No RAE") transform_dict['rand-crop'] = \ transforms.Compose([ transforms.CovertBGR(), transforms.Resize((origin_width)), transforms.RandomResizedCrop(scale=(ratio, 1), size=width), transforms.RandomHorizontalFlip(), transforms.ToTensor(), normalize, ]) else: print("RAE with", RAE, "mode") transform_dict['rand-crop'] = \ transforms.Compose([ transforms.CovertBGR(), transforms.Resize((origin_width)), transforms.RandomResizedCrop(scale=(ratio, 1), size=width), transforms.RandomHorizontalFlip(), transforms.ToTensor(), normalize, transforms.RandomErasing(mode=RAE, device='cpu') ]) transform_dict['center-crop'] = \ transforms.Compose([ transforms.CovertBGR(), transforms.Resize((origin_width)), transforms.CenterCrop(width), transforms.ToTensor(), normalize, ]) transform_dict['resize'] = \ transforms.Compose([ transforms.CovertBGR(), transforms.Resize((width)), transforms.ToTensor(), normalize, ]) return transform_dict
def __init__(self, root, train=True, test=True, transform=None): # Data loading code std_value = 1.0 / 255.0 mean_values = [104 / 255.0, 117 / 255.0, 128 / 255.0] if transform is None: transform = [ transforms.Compose([ transforms.CovertBGR(), transforms.Resize(256), transforms.RandomResizedCrop(227), transforms.RandomHorizontalFlip(), transforms.ToTensor(), transforms.Normalize(mean=mean_values, std=3 * [std_value]), ]), transforms.Compose([ transforms.CovertBGR(), transforms.Resize(256), transforms.CenterCrop(227), transforms.ToTensor(), transforms.Normalize(mean=mean_values, std=3 * [std_value]), ]) ] if root is None: root = 'DataSet/Products' traindir = os.path.join(root, 'train') testdir = os.path.join(root, 'test') if train: self.train = datasets.ImageFolder(traindir, transform[0]) if test: self.test = datasets.ImageFolder(testdir, transform[1])
def __init__(self, root, train=True, test=True, transform=None): # Data loading code mean_values = [0.485, 0.456, 0.406] std_values = [0.229, 0.224, 0.225] if transform is None: transform = [ transforms.Compose([ # transforms.CovertBGR(), transforms.Resize(256), transforms.RandomResizedCrop(224), transforms.RandomHorizontalFlip(), transforms.ToTensor(), transforms.Normalize(mean=mean_values, std=std_values), ]), transforms.Compose([ # transforms.CovertBGR(), transforms.Resize(256), transforms.CenterCrop(224), transforms.ToTensor(), transforms.Normalize(mean=mean_values, std=std_values), ]) ] if root is None: root = 'DataSet/Products' traindir = os.path.join(root, 'train') testdir = os.path.join(root, 'test') if train: self.train = datasets.ImageFolder(traindir, transform[0]) if test: self.test = datasets.ImageFolder(testdir, transform[1])
def __init__(self, root, train=True, test=True, transform=None): # Data loading code # std_values = [0.229, 0.224, 0.225] # mean_values = [104 / 255.0, 117 / 255.0, 128 / 255.0] normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) if transform is None: transform = [transforms.Compose([ # transforms.CovertBGR(), transforms.Resize(256), transforms.RandomResizedCrop(scale=(0.16, 1), size=224), transforms.RandomHorizontalFlip(), transforms.ToTensor(), normalize, ]), transforms.Compose([ # transforms.CovertBGR(), transforms.Resize(256), transforms.CenterCrop(224), transforms.ToTensor(), normalize, ])] if root is None: root = '/opt/intern/users/xunwang/DataSet//Products' traindir = os.path.join(root, 'train') testdir = os.path.join(root, 'test') if train: self.train = datasets.ImageFolder(traindir, transform[0]) if test: self.test = datasets.ImageFolder(testdir, transform[1])
def Generate_transform_Dict(origin_width=256, width=227, ratio=0.16): std_value = 1.0 / 255.0 normalize = transforms.Normalize(mean=[104 / 255.0, 117 / 255.0, 128 / 255.0], std= [1.0/255, 1.0/255, 1.0/255]) transform_dict = {} transform_dict['rand-crop'] = \ transforms.Compose([ transforms.CovertBGR(), transforms.Resize((origin_width)), transforms.RandomResizedCrop(scale=(ratio, 1), size=width), transforms.RandomHorizontalFlip(), transforms.ToTensor(), normalize, ]) transform_dict['center-crop'] = \ transforms.Compose([ transforms.CovertBGR(), transforms.Resize((origin_width)), transforms.CenterCrop(width), transforms.ToTensor(), normalize, ]) transform_dict['resize'] = \ transforms.Compose([ transforms.CovertBGR(), transforms.Resize((width)), transforms.ToTensor(), normalize, ]) return transform_dict