|
|
|
|
|
by utdemir
4361 days ago
|
|
I'm just today converted a 300ish lines of Python computer vision-machine learning script to MATLAB. Mind that I'm a long time Python programmer(4+ years), and just used MATLAB once or twice. But surprisingly, the resulting MATLAB code was fast, clean(cleaner than Python version) and also compatible with vast amaount of machine learning algorithms written in MATLAB(thats the very reason I had to switch to MATLAB). I always thought that Python can do everything cleaner and more elegant, but I'm sad that in MATLAB, its easier and cleaner. At least, now I don't need to import numpy, cv2, scipy, scikit-image, scikit-learn, pylab, matplotlib and ipython. Also, installing them was pretty hard too(specially scipy). |
|
However, there is one big (maybe a killing one) weakness of Matlab: it's proprietary. I used to work in a big company and need a Matlab license. The license costs 70k$, per year! It takes months to get the purchase approved. The biggest waste in choosing Matlab is the time to get it.
And because its proprietary, we'd have to KEEP buying for YEARS, otherwise our code won't run in the next project.
So we switched to Python in later projects, which does not require bureaucratic approval chains and provides code portability