thingsvision icon indicating copy to clipboard operation
thingsvision copied to clipboard

Published 20 hours ago • verified publisher icon ViCCo-Group

README improvement

Open LukasMut opened this issue 2 years ago • 2 comments

Improve readability of README file. Make text more concise and easier understandable?

LukasMut avatar Aug 23 '22 07:08 LukasMut

  • Include warning at the beginning that some original function calls mentioned in the paper are deprecated and that users should follow exclusively follow the README.

PhilippKaniuth avatar Sep 09 '22 12:09 PhilippKaniuth

  • Example snippets repeat alot of code, I think a general overview about what models are available and one example for how to call the extractor should be sufficient

andropar avatar Sep 12 '22 13:09 andropar

The readme very intuitive! I have a few suggestions, but none of them are critical imo.

  • In "About the project": Consider changing "bunch of images" to "a directory of images" or similar
  • In "Functionality": "extract features for the THINGS image database" is implicitly true from the previous two bullet points. I guess you aim to say something like "Conveniently access features for >26,000 images in the THINGS image database" to emphasize convenience?
  • "use HDF5 datasets": Sounds like a technical detail, maybe it's sufficient to mention this in the docs? The rest of the readme seems to be more of a high-level overview.
  • Under "Basic usage" -> "module_name": Maybe you could mention - in a short sentence - what you mean by "module"? Laymen might not be familiar?
  • In the example call for AlexNet: Is it possible to make it more visually distinct, which parts are code and which parts are console output? Otherwise, users might think they have to manually specify that big chunk of output produced by extractor.show_model()
  • Acknowledgements: I assume this section is not finalized yet? :)

oliver-contier avatar Nov 07 '22 14:11 oliver-contier

@oliver-contier could you add your review and make these changes (except for the final two points) directly in the PR? You can just go ahead and edit the current PR. Then @PhilippKaniuth can make additional changes to the docs and see how to address your second to the last point. I will take care of your last point and change the acknowledgments section into a contributions section (where we mention the people responsible for different parts of the package) as suggested in the thingsvision channel.

LukasMut avatar Nov 07 '22 16:11 LukasMut