If the images being entered are always in same condition, there’s no problem with the normal machine vision.
But even with the subtle changes in lighting, the detection rate is bound to vary greatly.
In such circumstances, there arises a need for people to put in and readjust the parameter.
However DL is designed to recognize and classify the object even in the midst of severe distortion of the image.
Since it recognizes the target at high speed, it’s also available in real-world applications.