Writing a review article is not plagiarism as long as you cite all sources correctly. If you don't, it's plagiarism.
Besides this, there are, at least in STM, limits regarding the amount of data that can be taken from a research paper due to copyright and permission handling, see the guidelines of the STM Association.
use up to three figures (including tables) from a journal article or
book chapter, but:
- not more than five figures from a whole book or journal issue/edition;
- not more than six figures from an annual journal volume; and
- not more than three figures from works published by a single publisher for an article, and not more than three figures from works published by a single publisher for a book chapter (and in total not more than thirty figures from a single publisher for re-publication in a book, including a multi-volume book, with different authors per chapter)
use single text extracts of less than 400 words from a journal article or book chapter, but
- not more than a total of 800 words from a whole book or journal issue/edition
Note, that in most cases you have to ask for permission or at least notify the publisher about the re-use of the material.
If you want to use more, you defenitely have to ask for permission and might be charged. But this is a copyright an not a plagiarism issue.