Almost every graph-cut paper I look at seems to have exactly the same pattern of monotonic growth in citations and then monotonic decline starting around 5 years ago:

enter image description here enter image description here enter image description here

For privacy I've cut the all author names out, but it seems that the people who published several graph-cut papers have the same (highly unusual) downward citation trend:

enter image description here enter image description here enter image description here enter image description here

These trends are highly uncommon in academia (if not completely uncommon except for in this unique case). What happened to the field of graph-cuts? Is some rising new interest such as Deep Learning provably better for the most popular application, in almost every way? Has something made graph-cuts obsolete for the most popular applications?

  • 1
    $\begingroup$ Maybe the the citation count of the recent years is not as high, because these new papers have not been disseminated enough? I mean, not everybody has read them, or else, some papers that are citing them are still work-in-progress and unpublished. $\endgroup$ – Robert Schwarz Oct 19 '19 at 8:15

Graph cuts were mainly used in computer vision, where since 2011 deep neural networks have taken over the field. The decline from 2015 on is attributable to a time delay in picking up neural networks.

Specifically, graph cuts were used for inferring maximum probable states in Markov Random Fields (MRF), with input costs coming from hand-tuned features. Solutions of these MRFs corresponded to some computer vision problem, be it segmentation, depth estimation in stereo, disparity estimation for optical flow etc. Currently, such tasks are solved to higher accuracy by neural networks without needing optimization, since their feature extraction part is much more sophisticated than previous pipelines which used graph cuts as a submodule.

Another less important issue is parallelization. Most efficient graph cut algorithms rely on combinatorial solvers that are difficult to parallelize, like the Boykov-Kolmogorov max-flow algorithm or the newer Maximum Flow by Incremental Breadth-First Search algorithm by Goldberg et al. Running them on graphics cards is rather difficult. Hence, even for very large-scale 3d-problems which were previously approached with graph cuts, this is less the case nowadays.

A third reason is that some people working on graph cuts have moved on and are now active in other fields, which are less cited than computer vision and machine learning.

| improve this answer | |
  • 1
    $\begingroup$ Thank you Paul! I see you are a brand new user, so probably joined this Stack Exchange specifically to answer this question. I'm curious how you came across this question? $\endgroup$ – Nike Dattani Mar 17 at 20:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.