data/realestate10k.py [18:69]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    def __init__(
        self, dataset, opts=None, num_views=2, seed=0, vectorize=False
    ):
        # Now go through the dataset

        self.imageset = np.loadtxt(
            opts.train_data_path + "/frames/%s/video_loc.txt" % "train",
            dtype=np.str,
        )

        if dataset == "train":
            self.imageset = self.imageset[0 : int(0.8 * self.imageset.shape[0])]
        else:
            self.imageset = self.imageset[int(0.8 * self.imageset.shape[0]) :]

        self.rng = np.random.RandomState(seed)
        self.base_file = opts.train_data_path

        self.num_views = num_views

        self.input_transform = Compose(
            [
                Resize((opts.W, opts.W)),
                ToTensor(),
                Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
            ]
        )

        self.offset = np.array(
            [[2, 0, -1], [0, -2, 1], [0, 0, -1]],  # Flip ys to match habitat
            dtype=np.float32,
        )  # Make z negative to match habitat (which assumes a negative z)

        self.dataset = "train"

        self.K = np.array(
            [
                [1.0, 0.0, 0.0, 0.0],
                [0, 1.0, 0.0, 0.0],
                [0.0, 0.0, 1.0, 0.0],
                [0.0, 0.0, 0.0, 1.0],
            ],
            dtype=np.float32,
        )

        self.invK = np.linalg.inv(self.K)

        self.ANGLE_THRESH = 5
        self.TRANS_THRESH = 0.15

    def __len__(self):
        return 2 ** 31
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



data/realestate10k.py [244:295]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    def __init__(
        self, dataset, opts=None, num_views=2, seed=0, vectorize=False
    ):
        # Now go through the dataset

        self.imageset = np.loadtxt(
            opts.train_data_path + "/frames/%s/video_loc.txt" % "train",
            dtype=np.str,
        )

        if dataset == "train":
            self.imageset = self.imageset[0 : int(0.8 * self.imageset.shape[0])]
        else:
            self.imageset = self.imageset[int(0.8 * self.imageset.shape[0]) :]

        self.rng = np.random.RandomState(seed)
        self.base_file = opts.train_data_path

        self.num_views = num_views

        self.input_transform = Compose(
            [
                Resize((opts.W, opts.W)),
                ToTensor(),
                Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
            ]
        )

        self.offset = np.array(
            [[2, 0, -1], [0, -2, 1], [0, 0, -1]],  # Flip ys to match habitat
            dtype=np.float32,
        )  # Make z negative to match habitat (which assumes a negative z)

        self.dataset = "train"

        self.K = np.array(
            [
                [1.0, 0.0, 0.0, 0.0],
                [0, 1.0, 0.0, 0.0],
                [0.0, 0.0, 1.0, 0.0],
                [0.0, 0.0, 0.0, 1.0],
            ],
            dtype=np.float32,
        )

        self.invK = np.linalg.inv(self.K)

        self.ANGLE_THRESH = 5
        self.TRANS_THRESH = 0.15

    def __len__(self):
        return 2 ** 31
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -