What is np.where(landmarks[:, a] == -1, -1, landmarks[:, a]) for numpy value

Recently, I was learning and re using tensorflow to copy yolov5, and I found that there were some np Where (landmarks[:, a] = -1, -1, landmarks[:, a]) operation. I think this operation is meaningless. They seem to be similar to if (true). What is the function of this thing?

can you post a little bit more detail? maybe some code snippet?
that would be easier for people to share some insights