fengyang0317 / unsupervised_captioning

Code for Unsupervised Image Captioning
MIT License
215 stars 51 forks source link

Suggestions #1

Closed gautamsreekumar closed 5 years ago

gautamsreekumar commented 5 years ago

Some corrections and suggestions about the repo.

  1. The installation instructions are confusing - whether workspace is a directory inside unsupervised-captioning or the other way. "First we refer to requirements.txt and then we clone the repo" seems wrong.
  2. In preprocessing/crawl_descriptions.py, obj = re.search('data-max="(\d*)"', page) should be obj = re.search('max="(\d*)"', page). I went through the HTML file and this is what I found (as of 8 April, 2019. The format could change later).
  3. preprocessing/extract_descriptions.py imports config.py which is located in the root dir of the repo. Shouldn't this file be inside preprocessing dir. Also, the workspace dir needs maybe modified accordingly. Same with misc_fn.py
  4. An error may occur while running preprocessing/extract_descriptions.py. This is a solution I found in GitHub -> answer
fengyang0317 commented 5 years ago

I assume that you are executing the commands under "Requirements" sequentially. The unsupervised_captioning is inside workspace directory.

gautamsreekumar commented 5 years ago

@fengyang0317 Sorry, I was not aware I had posted the comments. I have updated my comments. Please check it again. So what I was saying is that requirements.txt is a file inside the repo. But the repo is cloned after workspace directory is created.

fengyang0317 commented 5 years ago

Thanks for your suggestion. I was not aware of this issue. I have updated the Requirements part.

fresh382227905 commented 4 years ago

thanks, it does work.