Пример #1
0
    def __init__(self, sequence=5):
        self.scene = glob(PATH)
        self.scene.sort()

        self.sequence = sequence
        self.list = []
        self.list_coordinate = []
        self.all_data = []

        for i, j in zip(self.scene, list(range(len(self.scene)))):
            self.out = []
            data_split = []
            print(i, "----------------------------")
            print(j)
            sub_path = os.path.join(i, "annotations4.txt")  # change _ to 4
            data = np.genfromtxt(sub_path, delimiter=' ')
            # Center coordinates from Bounding Box
            x_lim1, y_lim1 = data[:, 1], data[:, 2]
            x_lim2, y_lim2 = data[:, 3], data[:, 4]
            x = (x_lim1 + x_lim2) // 2
            y = (y_lim1 + y_lim2) // 2

            # Normalization
            x, y = train(x, y, j)

            data[:, 1], data[:, 2] = x, y
            # all target in the scene
            ID = np.unique(data[:, 0][-1])
            for j in range(0, int(ID[0])):
                trajectory = data[data[:, 0] == j, :]
                # ID,x,y,frame,attribute
                trajectory = trajectory[:, [0, 1, 2, 5, 9]]
                # extract attribute
                trajectory = Attribute(trajectory)
                if len(trajectory) > 0:
                    data_split.append(trajectory)
            for split in range(len(data_split)):
                data_split_coor = data_split[split]
                if data_split_coor.shape[0] > self.sequence + 1:
                    self.out.append(data_split_coor[:,
                                                    [1, 2, 4, 5, 6, 7, 8, 9]])
                    self.all_data.append(
                        data_split_coor[:, [1, 2, 4, 5, 6, 7, 8, 9]])

            self.list_coordinate.append(self.out)
            self.list.append(len(self.out))
Пример #2
0
    def __init__(self, sequence=10):
        self.path = glob(PATH)
        self.path.sort()

        self.sequence = sequence
        self.list = []
        self.list_coordinate = []
        self.all_data = []

        for i, j in zip(self.path, range(len(self.path))):
            print i, "----------------------------"
            print j
            self.out = []
            trajectory_split = []

            sub_path = os.path.join(i, "annotations_.txt")
            data = np.genfromtxt(sub_path, delimiter=' ')

            # Center coordinates from Bounding Box
            x_lim1, y_lim1 = data[:, 1], data[:, 2]
            x_lim2, y_lim2 = data[:, 3], data[:, 4]
            x = (x_lim1 + x_lim2) / 2
            y = (y_lim1 + y_lim2) / 2

            # Normalization
            x, y = train(x, y, j)

            data[:, 1], data[:, 2] = x, y
            # all target in the scene
            ID = np.unique(data[:, 0][-1])
            for j in range(0, int(ID[0])):
                trajectory = data[data[:, 0] == j, :]
                # ID,x,y,frame
                trajectory = trajectory[:, [0, 1, 2, 5]]
                if len(trajectory) > 0:
                    trajectory_split.append(trajectory)

            for split in xrange(len(trajectory_split)):
                data_split = trajectory_split[split]
                if data_split.shape[0] > self.sequence + 1:
                    self.out.append(data_split[:, [1, 2]])
                    self.all_data.append(data_split[:, [1, 2]])

            self.list_coordinate.append(self.out)
            self.list.append(len(self.out))