Visual Relationship Detection using Scene Graphs: A Survey (Submission)

Mathpix / spectra-review-paper-competition

Competition for best expository article on cutting-edge ML research

https://spectra.pub

56 stars 40 forks source link

Visual Relationship Detection using Scene Graphs: A Survey (Submission) #29

Closed aniket-agarwal1999 closed 3 years ago

aniket-agarwal1999 commented 3 years ago

Submission of a survey on scene graphs for this competition. Thanks :)

th-anderson commented 3 years ago

This review article has already been published:

https://arxiv.org/abs/2005.08045

aniket-agarwal1999 commented 3 years ago

That is our "Arxiv" paper as well. This submission is a modified version of that and has some updations.

aniket-agarwal1999 commented 3 years ago

@nicodjimenez can u please see the PR because I suppose the deadline is really close. Thanks :)

nicodjimenez commented 3 years ago

Merging, thanks for the submission!

Can you elaborate more on what you updated since that Arxiv submission?

nicodjimenez commented 3 years ago

Btw submission is live at:

https://spectra.pub/ml/visual-relationship-scene-graph

aniket-agarwal1999 commented 3 years ago

Yeah sure, in this submission we have trimmed many of the unecessary stuff to cut the review a bit short and also removed description about some of the really old techniques as well which served really less to the purpose of giving a brief and wholesome gist of the subject. Apart from this, we have mainly updated in the Future Directions section a bit to refine it since we believed there were many other directions in this field that we did not know when finalizing the arxiv submission. Also updated the summarization of some of the methods for better understanding of the reader and same in the Applications section as well.

aniket-agarwal1999 commented 3 years ago

Hi @nicodjimenez , our submission seems to have been removed. Is there a specific reason for it?

nicodjimenez commented 3 years ago

Ah you're right! Thanks for bringing this to our attention. In any case, it's now fixed. It was caused by one of the articles having broken frontmatter.