datasets.py

import os
import json

from torchvision import datasets, transforms
from torch.utils.data import Dataset
from torchvision.datasets.folder import ImageFolder, default_loader

from timm.data.constants import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD
from timm.data import create_transform
""" Stanford Cars (Car) Dataset
Created: Nov 15,2019 - Yuchong Gu
Revised: Nov 15,2019 - Yuchong Gu
"""
import os
# import pdb
from PIL import Image
import pickle
# from scipy.io import loadmat



class CarsDataset(Dataset):
    """
    # Description:
        Dataset for retrieving Stanford Cars images and labels
    # Member Functions:
        __init__(self, phase, resize):  initializes a dataset
            phase:                      a string in ['train', 'val', 'test']
            resize:                     output shape/size of an image
        __getitem__(self, item):        returns an image
            item:                       the idex of image in the whole dataset
        __len__(self):                  returns the length of dataset
    """

    def __init__(self, root, train=True, transform=None):
        self.root = root
        self.phase = 'train' if train else 'test'
        # self.resize = resize
        self.num_classes = 196

        self.images = []
        self.labels = []

        list_path = os.path.join(root, 'cars_anno.pkl')

        list_mat = pickle.load(open(list_path, 'rb'))
        num_inst = len(list_mat['annotations']['relative_im_path'][0])
        for i in range(num_inst):
            if self.phase == 'train' and list_mat['annotations']['test'][0][i].item() == 0:
                path = list_mat['annotations']['relative_im_path'][0][i].item()
                label = list_mat['annotations']['class'][0][i].item()
                self.images.append(path)
                self.labels.append(label)
            elif self.phase != 'train' and list_mat['annotations']['test'][0][i].item() == 1:
                path = list_mat['annotations']['relative_im_path'][0][i].item()
                label = list_mat['annotations']['class'][0][i].item()
                self.images.append(path)
                self.labels.append(label)

        print('Car Dataset with {} instances for {} phase'.format(len(self.images), self.phase))

        # transform
        self.transform = transform

    def __getitem__(self, item):
        # image
        image = Image.open(os.path.join(self.root, self.images[item])).convert('RGB')  # (C, H, W)
        image = self.transform(image)

        # return image and label
        return image, self.labels[item] - 1  # count begin from zero

    def __len__(self):
        return len(self.images)


class INatDataset(ImageFolder):
    def __init__(self, root, train=True, year=2018, transform=None, target_transform=None,
                 category='name', loader=default_loader):
        self.transform = transform
        self.loader = loader
        self.target_transform = target_transform
        self.year = year
        # assert category in ['kingdom','phylum','class','order','supercategory','family','genus','name']
        path_json = os.path.join(root, f'{"train" if train else "val"}{year}.json')
        with open(path_json) as json_file:
            data = json.load(json_file)

        with open(os.path.join(root, 'categories.json')) as json_file:
            data_catg = json.load(json_file)

        path_json_for_targeter = os.path.join(root, f"train{year}.json")

        with open(path_json_for_targeter) as json_file:
            data_for_targeter = json.load(json_file)

        targeter = {}
        indexer = 0
        for elem in data_for_targeter['annotations']:
            king = []
            king.append(data_catg[int(elem['category_id'])][category])
            if king[0] not in targeter.keys():
                targeter[king[0]] = indexer
                indexer += 1
        self.nb_classes = len(targeter)

        self.samples = []
        for elem in data['images']:
            cut = elem['file_name'].split('/')
            target_current = int(cut[2])
            path_current = os.path.join(root, cut[0], cut[2], cut[3])

            categors = data_catg[target_current]
            target_current_true = targeter[categors[category]]
            self.samples.append((path_current, target_current_true))

    # __getitem__ and __len__ inherited from ImageFolder


def build_dataset(is_train, args, infer_no_resize=False):
    transform = build_transform(is_train, args, infer_no_resize)

    if args.data_set == 'CIFAR':
        dataset = datasets.CIFAR100(args.data_path, train=is_train, transform=transform, download=True)
        nb_classes = 100
    elif args.data_set == 'CIFAR10':
        dataset = datasets.CIFAR10(args.data_path, train=is_train, transform=transform, download=True)
        nb_classes = 10
    elif args.data_set == 'CARS':
        dataset = CarsDataset(args.data_path, train=is_train, transform=transform)
        nb_classes = 196
    elif args.data_set == 'FLOWERS':
        root = os.path.join(args.data_path, 'train' if is_train else 'test')
        dataset = datasets.ImageFolder(root, transform=transform)
        nb_classes = 102
    elif args.data_set == 'IMNET':
        root = os.path.join(args.data_path, 'train' if is_train else 'val')
        dataset = datasets.ImageFolder(root, transform=transform)
        nb_classes = 1000
    elif args.data_set == 'INAT':
        dataset = INatDataset(args.data_path, train=is_train, year=2018,
                              category=args.inat_category, transform=transform)
        nb_classes = dataset.nb_classes
    elif args.data_set == 'INAT19':
        dataset = INatDataset(args.data_path, train=is_train, year=2019,
                              category=args.inat_category, transform=transform)
        nb_classes = dataset.nb_classes

    return dataset, nb_classes


def build_transform(is_train, args, infer_no_resize=False):
    if hasattr(args, 'arch'):
        if 'cait' in args.arch and not is_train:
            print('# using cait eval transform')
            transformations = {}
            transformations= transforms.Compose(
                [transforms.Resize(args.input_size, interpolation=3),
                transforms.CenterCrop(args.input_size),
                transforms.ToTensor(),
                transforms.Normalize(IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD)])
            return transformations
    
    if infer_no_resize:
        print('# using cait eval transform')
        transformations = {}
        transformations= transforms.Compose(
            [transforms.Resize(args.input_size, interpolation=3),
            transforms.CenterCrop(args.input_size),
            transforms.ToTensor(),
            transforms.Normalize(IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD)])
        return transformations

    resize_im = args.input_size > 32
    if is_train:
        # this should always dispatch to transforms_imagenet_train
        transform = create_transform(
            input_size=args.input_size,
            is_training=True,
            color_jitter=args.color_jitter,
            auto_augment=args.aa,
            interpolation=args.train_interpolation,
            re_prob=args.reprob,
            re_mode=args.remode,
            re_count=args.recount,
        )
        if not resize_im:
            # replace RandomResizedCropAndInterpolation with
            # RandomCrop
            transform.transforms[0] = transforms.RandomCrop(
                args.input_size, padding=4)
        return transform

    t = []
    if resize_im:
        size = int((256 / 224) * args.input_size)
        t.append(
            transforms.Resize(size, interpolation=3),  # to maintain same ratio w.r.t. 224 images
        )
        t.append(transforms.CenterCrop(args.input_size))

    t.append(transforms.ToTensor())
    t.append(transforms.Normalize(IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD))
    return transforms.Compose(t)