Yep, this can be considered pseudo-polynomial time.
The abstract says "computes the prime factorisation of a positive integer N", so it's not your fault you were confused. The abstract should be corrected.
There’s nothing wrong with the abstract as written. It makes no claim that the algorithm is polynomial-time.
The abstract is correct. A positive integer N is log(N) bits long. If X = log(N), than a algorithm that runs in N^(1/5) will run in 2^(X/5) time.