Vertical flip is a basic image data augmentation method that flips an image 180 degrees along the vertical central axis to generate a new upside-down image sample. It does not change the core features of the image, and is often used to expand the training set and improve the model's adaptability to vertical changes of targets.
