There are many issues with the 'install'
1.) The readme says that as of version 3.0 scipy and peakutils are no longer needed
2.) the current readme says to only use 'sudo apt install python34-opencv' of course that will not work as you must explicitly type 'pip install opencv-python'
There are dozens of 'run' problems, it appears that nobody has tested or tried to build; This is common on +5 year old software, the python has moved on; There are pip tools that generate a 'requirements.txt' file that specify exactly what version of the tools 'numpy' to be used, as this is the core of the problem that the 'code' is using obsolete 'types'; The problem here is chicken and the egg, we don't have a 'working' version so we can't generate the requirements.txt; Only the system that has the working version can generate;
This comment about install is the requirements ( with no version info ) for the version 1.0, of little use to version 3.0 which has been updated, but still now this update is 3 years old;
I Haven't even tried to run on a 'pi' yet ( I have dozens of rasperry's ) I just tried 3 different LINUX systems and all the same problem, the program crashes; Mostly because there is no 'error throw' code, the program trys to read the camera and config a calibration on first run, and if it doesn't find a 'spectrogram' it just crashes, there should be graceful 'throw' code so that there is no crash;
...
For 'working' alternatives in 2024, I would suggest thermino(Italian), spectrogryph( German), or https://spectralworkbench.org/dashboard ( still working, but closed for lack of funding), the latter is a 'browser' solution but works standalone, and lets you capture any spectogram then process it using there tools;
,,,
It appears that the README needs to be completely rewritten; OP-AUTHOR should find somebody to help them update the software build/run tools so that anybody can automatically run this software without days of frustration; If that fails, then the OP-AUTHOR can build a 'python-binary' on his system and just post that on the GitHub so the user doesn't have to maintain obsolete librarys in order to build; The problem is here, the program uses obsolete library's but we don't know what they are, and it would take a long time trial and error to find a working combination where there are 1,000's of library's with 100's of version's
There are many issues with the 'install' 1.) The readme says that as of version 3.0 scipy and peakutils are no longer needed 2.) the current readme says to only use 'sudo apt install python34-opencv' of course that will not work as you must explicitly type 'pip install opencv-python'
There are dozens of 'run' problems, it appears that nobody has tested or tried to build; This is common on +5 year old software, the python has moved on; There are pip tools that generate a 'requirements.txt' file that specify exactly what version of the tools 'numpy' to be used, as this is the core of the problem that the 'code' is using obsolete 'types'; The problem here is chicken and the egg, we don't have a 'working' version so we can't generate the requirements.txt; Only the system that has the working version can generate;
This comment about install is the requirements ( with no version info ) for the version 1.0, of little use to version 3.0 which has been updated, but still now this update is 3 years old;
I Haven't even tried to run on a 'pi' yet ( I have dozens of rasperry's ) I just tried 3 different LINUX systems and all the same problem, the program crashes; Mostly because there is no 'error throw' code, the program trys to read the camera and config a calibration on first run, and if it doesn't find a 'spectrogram' it just crashes, there should be graceful 'throw' code so that there is no crash;
...
For 'working' alternatives in 2024, I would suggest thermino(Italian), spectrogryph( German), or https://spectralworkbench.org/dashboard ( still working, but closed for lack of funding), the latter is a 'browser' solution but works standalone, and lets you capture any spectogram then process it using there tools;
,,,
It appears that the README needs to be completely rewritten; OP-AUTHOR should find somebody to help them update the software build/run tools so that anybody can automatically run this software without days of frustration; If that fails, then the OP-AUTHOR can build a 'python-binary' on his system and just post that on the GitHub so the user doesn't have to maintain obsolete librarys in order to build; The problem is here, the program uses obsolete library's but we don't know what they are, and it would take a long time trial and error to find a working combination where there are 1,000's of library's with 100's of version's