What are the alternatives to convert .m files to Python, besides doing it manually? I need them to run on Raspberry Pi

Question

I have been given a matlab file (rather files) and have been asked to convert it to a code compatible for Raspberry Pi. The problem is that Raspberry Pi doesn't support Matlab and manual conversion of Matlab to Python is one another hell of a task.

I am giving a sample code for example that I need to convert (This is a huge code about a modified ACF Detector).

    %%%%%%%%%%%%%%%%%%%%%%%%%%% WELCOME %%%%%%%%%%%%%%%%%%%%%%%%%%%

%SETUP
PMTroot = '/home/lisa/Downloads/toolbox-master'; %SET LOCATION OF PIOTR'S TOOLBOX
dataDir =''; %SET LOCATION OF CALTECH DATA

addpath(genpath(PMTroot)); %YOU MAY NEED TO PLACE acfDetect_modified.m where acfDetect.m is

%%
octupSET = 1;
opts = acfTrain;
opts.modelDs=[50 20.5].*(2^octupSET); opts.modelDsPad=[64 32].*(2^octupSET);

%DEFAULTS
opts.pPyramid.pChns.pColor.smooth=0; opts.nWeak=[64 256 1024 4096];
opts.pBoost.discrete=0; opts.pPyramid.pChns.pGradHist.softBin=1;
opts.pPyramid.nOctUp=octupSET;
opts.pPyramid.minDs = opts.modelDs;
opts.pPyramid.pad = ceil((opts.modelDsPad-opts.modelDs)/opts.pPyramid.pChns.shrink/2)*opts.pPyramid.pChns.shrink;
opts.seed = 0;
pLoad={'lbls',{'person'},'ilbls',{'people'},'squarify',{3,.41}};
%If doing flip augmentation, should use %pLoad={'lbls',{'person'},'ilbls',{'people'},'squarify',[]};
opts.pLoad = [pLoad 'hRng',[opts.modelDs(1)./2^opts.pPyramid.nOctUp inf], 'vRng',[1 1] ];
opts.winsSave = 0;
opts.pPyramid.pChns.shrink= 4; %2^(octupSET+1);


%SET NAME
opts.name=['Caltech'];


%OUR CHANGES
opts.pBoost.pTree.maxDepth=6;
opts.pBoost.pTree.fracFtrs=1; %1/8; %1/16; %Depending on how much time you have for training
opts.nNeg=100000;
opts.nAccNeg=2*opts.nNeg;


%JITTER POSITIVE SAMPLES.
%May benefit from more, up to you. sclsRange = [1.11:0.01:1.15] or sclsRange = [1.05:0.01:1.15]
%helps ACF a bit
sclsRange = 1.1;
sclsArray = [1 1];
for j=sclsRange
    sclsArray = [sclsArray; j 1; 1 j; j j];
end
opts.pJitter=struct('flip',0,'scls',sclsArray);

%OPTION 1: Obtain new annotations from https://www.mpi-inf.mpg.de/departments/computer-vision-and-multimodal-computing/research/people-detection-pose-estimation-and-tracking/how-far-are-we-from-solving-pedestrian-detection/
bnewanno = 0;
if(bnewanno)
    opts.posGtDir =[ dataDir '/anno_train10x_alignedby_RotatedFilters'];
    opts.posImgDir=[dataDir  '/train_03_clean' '/images_clean'];
    %delete([opts.posImgDir '/' 'Thumbs.db'])
    %delete([opts.posGtDir '/' 'Thumbs.db'])
else
    %We sample every 4th video frame
    opts.posGtDir=[dataDir '/train04/annotations/' ];
    opts.posImgDir=[dataDir '/train04/images/' ];
end

%OPTION 2: LDCF84 OR ACF
bldcf = 0;
if(bldcf)
    opts.filters=[2 4];
end

detector=acfTrain_modified(opts);
%%
%Testing
pLoad={'lbls',{'person'},'ilbls',{'people'},'squarify',{3,.41}};
pModify=struct('cascThr',-1,'cascCal',.025);

if(bnewanno==0)
    imgDir = [dataDir 'test/images'];
    gtDir = [dataDir 'test/annotations'];
else
    %New annotations
    imgDir = [dataDir 'test/images'];
    gtDir = [dataDir 'test/anno_test_1xnew'];
end

bontrain = 0; %For context model
if(bontrain)
    if(bneweval==0)
        samp = '04';
        gtDir=[dataDir '/train' samp '/annotations'];
        imgDir=[dataDir '/train' samp '/images'];
    else
        gtDir=[ dataDir '/anno_train10x_alignedby_RotatedFilters'];
        imgDir=[ dataDir '/train_03_clean' '/images_clean'];
    end

    [miss,roc,gt,dt,pr]=acfTest_modified('name',name,'imgDir',imgDir,...
        'gtDir',gtDir,'pLoad',[pLoad, 'hRng',[50 inf],...
        'vRng',[.65 1],'xRng',[5 635],'yRng',[5 475]],...
        'pModify',pModify,'reapply',0,'show',0,'outname',[name 'approx0ontrainDets.txt']);
else

    [miss,roc,gt,dt,pr]=acfTest_modified('name',name,'imgDir',imgDir,...
        'gtDir',gtDir,'pLoad',[pLoad, 'hRng',[50 inf],...
        'vRng',[.65 1],'xRng',[5 635],'yRng',[5 475]],...
        'pModify',pModify,'reapply',0,'show',0,'outname',[name 'approx0Dets.txt']);
end

%%

I wanted help on whether what tool can be applied to convert it to Raspberry Pi compatible code (Python preferably). SMOP and OMPC don't seem to work very well since they simply try to convert to Python. In this process, they however, miss certain MATLAB functions which are copied to the converted code as such and hence not very useful.

I saw the answer on A tool to convert MATLAB code to Python, but I am not able to figure out how should the interfacing be done. I am not very proficient in MATLAB, so will this mean I have to understand everything about not only about this code, but also of NumPy, matplotlib(if I intend to use them). Is there an alternative?