search

zegami / image-similarity-clustering

This project allows images to be automatically grouped into like clusters using a combination of machine learning techniques.
MIT License
236 stars 50 forks source link

file formats #7

Open dnnguifever opened 6 years ago

dnnguifever commented 6 years ago

Hello Roger,

I've just come across this, and it looks quite impressive. I'd like to try it out. Just a couple questions.

In your description for extract.py, you write:

The results are saved to a tab separated file postfixed with '_features.

And then, later you say that tsne.py "takes a comma separate list of values".

So, I am new to the field, forgive me. Is the tab-separate output file of extract.py supposed to be the comm-separated file of tsne.py?

The output of tsne.py is "tab separated"; these tab-separated values are essentially the X and Y values that can be plotted?

Thanks! Much appreciated.

shweta112 commented 5 years ago

@dnnguifever I think the "comma separate list of values" refers to the features which are stored as comma separated within the tab separated file postfixed with '_features'. This file is the input to tsne.py which ultimately outputs another tab separated file that you can use to create a plot.

You probably figured it out yourself or don't need the answer anymore; just putting it here in case others have the same doubt.