python - 如何创建一个与另一个元组形状相同的元组

Question

我正在尝试增加数据以解决 CCN 问题。我有一个包含 3 个图像(中、左、右)的 csv，并且所有三个图像的转向角度都相同。 (我使用了 Udacity 自动驾驶汽车模拟器)。我想增加我的数据，所以我尝试对我的图像进行增强。我需要创建一个具有如下形状的元组:

image_path = data[["center", "left", "right"]].values

但在此之前，我必须增强我的数据。这是我的代码:

steerings = data["steering"].values
steerings = list(steerings)
center = data["center"].values
left = data["left"].values
right = data["right"].values
center = list(center)
new_center = []
left = list(left)
new_left = []
right = list(right)
new_right = []
new_steerings = []
for index in range(len(center)):
    new_center_img = augment_image(center[index], steerings[index])[0]
    new_center.append(new_center_img)
    new_left_img = augment_image(left[index], steerings[index])[0]
    new_left.append(new_left_img)
    new_right_img = augment_image(right[index], steerings[index])[0]
    new_right.append(new_right_img)
    new_steerings.append(steerings[index])
center.extend(new_center)
left.extend(new_left)
right.extend(new_right)
steerings.extend(new_steerings)

image_path = tuple(center), tuple(left), tuple(right)
steerings = tuple(steerings)


return train_test_split(image_path, steerings, test_size=0.2, random_state=1)

但是:

image_path = tuple(center), tuple(left), tuple(right)

不等于:

image_path = data[["center", "left", "right"]].values

当我返回图像路径时，如何获得相同的东西，但由于增强而具有双图像？

Answer 1

如果我理解正确，我认为你只需要这样做:

image_path = np.stack([center, left, right], axis=1)

其中 np 是 NumPy。