Error during installation

[yuerout]

Hi! I'm trying to install ibeis with pip install ibeis but I got this error:

Collecting pyflann-ibeis
  Using cached pyflann_ibeis-2.0.0.tar.gz (168 kB)
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error

  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [8 lines of output]
      running egg_info
      creating /tmp/pip-pip-egg-info-mlh0hi_4/pyflann_ibeis.egg-info
      writing /tmp/pip-pip-egg-info-mlh0hi_4/pyflann_ibeis.egg-info/PKG-INFO
      writing dependency_links to /tmp/pip-pip-egg-info-mlh0hi_4/pyflann_ibeis.egg-info/dependency_links.txt
      writing requirements to /tmp/pip-pip-egg-info-mlh0hi_4/pyflann_ibeis.egg-info/requires.txt
      writing top-level names to /tmp/pip-pip-egg-info-mlh0hi_4/pyflann_ibeis.egg-info/top_level.txt
      writing manifest file '/tmp/pip-pip-egg-info-mlh0hi_4/pyflann_ibeis.egg-info/SOURCES.txt'
      error: package directory 'pyflann_ibeis/lib' does not exist
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

How can I fix this?

[Erotemic]

What operating system and Python version are you using? I'm hoping to spend some time updating the release packages in the next month, but in the meantime Ubuntu with Python 3.8 should work.

[yuerout]

I used Ubuntu with python 3.9. Fixed this by creating a conda environment with python=3.6. Thank you! Also is there a more comprehensive user guide to the Ibeis GUI? I would like to use the scores computed by Ibeis for some classification, but I'm a bit confused on how to use the GUI now.

[Erotemic]

No, there is no comprehensive guide. (Although feel free to make one and contribute it!)

The README is the best I have. The "Advanced ID Interface" is IMO the easiest way to use the program, but it has a learning curve. The main idea is you define what your "query" and "database" sets are as well as algorithm parameters (which usually don't need to change), and also review parameters (which you may want to tweak).

I will caution you about using the scores for any type of classification. The one-vs-many scores are highly volatile and non-comparable. They simply don't provide a separable distribution that can be used to make any type of automated decision. The manual review is necessary. However, within a single query-to-database result, the correct results do tend to show up at the top of the ranked list more often than not.

Part of the final stage of my research was a verification classifier (we called VAMP), which tried to learn a random forest on one-verus-one scores to build such an automatic classifier. That did work reasonably well, but unfortunately it never made it into the IBEIS GUI proper. However, the WBIA (wildbook image analysis) module does have the method integrated IIRC. So I strongly recommend checking out the wildbook platform and using that instead. You will need to reach out to wildme to see if they can support your use case.

[Erotemic]

This should now be fixed with Python 3.9. IBEIS and all of its dependencies just got some much needed maintenance and new versions on pypi.