'Point d'accumulation' runs into the same "problem" as 'valeur d'adhérence' because as defined for a set, an accumulation point of a subset A of a metric space M is a point a of M for which for all e>0, [itex](B_{\epsilon}(a)\backslash \{a\})\cap A\neq \emptyset[/itex] (i.e. every open ball centered on a contains points of A other than a).
So even if we apply this concept to the image of the sequence, every accumulation point is a cluster point but not every cluster point is an accumulation point. For instance consider the constant sequence 1,1,1,... 1 is a cluster point but the set of all accumulation points of the image, {1}, is void.