A lot of work regarding Graph Neural Networks require fixed number of nodes. In the case of image processing using graph, the image representation is often super pixels (like in this work https://arxiv.org/pdf/2002.05544.pdf)
The other example is in this work (https://arxiv.org/pdf/1611.08402.pdf) with MNIST Superpixels which also ends u having same number of superpixels.
A lot of algorithms including recommended SLIC doesn’t follow this and generate different number of superpixles.
How to handle this? Are there some implementations which generate same number of superpixels in place or there is other mechanism used?