In the first sentence, the important observation is that the sheep crossed the road. What was seen is the crossing. The fact that I've seen sheep is (mostly) irrelevant.
In the second sentence, the relevant is that the police spotted him. What was seen is him. The last part of the sentence merely specifies the conditions, but they are considered less relevant. It could be rephrased like this:
La police l'a aperçu. Il était en train d'entrer dans un laboratoire médical.
The following is also correct and slightly different:
La police l'a aperçu entrer dans un laboratoire médical.
The focus here is on “he entered the lab”. It lessens the importance of the fact he was spotted. It might even suggest that the trespass is the (or part of the) transgression.