I don’t count having no visual indication of the object as “tracking” it, if we’re talking semantics. One frame could equal an even faster speed than what it would minimally take to cross the entire width of the image at some trajectory vector. For other vectors, it could be (much) less (like not passing straight through the image from on side to the opposite side, e.g.).

It’s important to not hang too hard on this as the escape speed is dependent on air resistance, or rather lack thereof. Those escape speed numbers are defined along with the assumption of zero air resistance or other forces acting on the object.