petrhruby97 / learning_minimal Goto Github PK

Dear Petr Hrubý,

Congratulations for your great work. I am glad to see source code derived from MINUS (https://github.com/rfabbri/minus) being widely useful. However, there are licensing issues that need to be addressed.

MINUS has an explicit LICENSE file in its sourcecode: https://github.com/rfabbri/minus/blob/master/LICENSE

Which mentions explicit copyright and conditions to be met:

MINUS - MInimial problem NUmerical continuation Solver
Released under the BSD 2-Clause License:

Copyright (c) 2019 Ricardo Fabbri, Anton Leykin and Timothy Duff

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

1. Redistributions of source code must retain the above copyright notice, this
   list of conditions and the following disclaimer.
   ...

Your path tracking source code in:

Derives from MINUS source code in:
https://github.com/rfabbri/minus/blob/f01e1b72eef1a15184998d25d30f6b54564a9c17/minus/minus.hxx#L26

There is a number of other places that the code derives from MINUS, but the license conditions have not been met. For instance:

Your source code in:

Derives from MINUS source code in:
https://github.com/rfabbri/minus/blob/master/minus/internal-util.h

Your source code in:

Derives from MINUS source code in (different evaluator, but same code style):
https://github.com/rfabbri/minus/blob/master/minus/chicago14a.hxx

The specific programming practice used in MINUS was the outcome of my efforts to optimize Homotopy Continuation. Therefore, the resulting style and C/C++ constructs are unique and traceable.

Please realize that I admire your work, I am just reminding you that the LICENCE conditions be met. In addition to the aforementioned legal LICENCING terms, I would like remind you that it is good scholarly practice to kindly credit the sources of your work in the README.

Thank you for your work, I will for sure use it myself and cite it.
Ricardo Fabbri

Hi, I very much enjoyed your paper and has been working on a similar project. I have some questions regarding your classifier. If each possible anchor ID is set as an expected label, does this mean that this MLP classifier chooses from a label set, say 1000 starting anchors, and for each new problem, select, say 20 valid anchors from them? This would leave the possible label space with approximately 2^1000 choices? Wouldn't this be too large or did I understand your paper incorrectly? Thank you very much!

At least no novel&fancy network this year

Hi,

Could you please tell me, what is the entry point for the user? Let's say for 5pt thing, Is it https://github.com/petrhruby97/learning_minimal/blob/main/5p2v/src/ransac.hxx#L98 ?

Very nice paper. I enjoyed discussing this work with you at CVPR, and I'm keen to be able to reproduce your results. If anyone would like to follow my attempt to replicate this work, I have created a branch on my fork where I will post extended readme information, instructions, requirments.txt, some light code clean-up, etc.

There is an undefined variable on line 15 of sample_data.py. This appears to affect both the 4p3v and 5p2v versions of the code, which contain identical versions of sample_data.py.

Here is the traceback:

Traceback (most recent call last):
  File "/home/james/Projects/learning_minimal/4p3v/sample_data.py", line 47, in <module>
    generate(input_folder, output_file, goal_samples)
  File "/home/james/Projects/learning_minimal/4p3v/sample_data.py", line 15, in generate
    print("generating " + name)
NameError: name 'name' is not defined