AUTO-MULTIPLE-CHOI(1)
π
AUTO-MULTIPLE-CHOI(1) β like, the auto-multiple-choice-analyse page, bestie
okay so real talk, this is the manual page for Auto Multiple Choice, and this specific one is auto-multiple-choice-analyse. Sounds fancy, is fancy. Let's get into it β¨
β¨ εε (NAME) β¨
auto-multiple-choice-analyse β literally the thing that goes brrr on your scanned AMC multiple-choice answer sheets and automatically recognizes the marks π― no manual counting, no tears, periodt.
πΈ ζ¦θ¦ (SYNOPSIS) πΈ
auto-multiple-choice analyse --projet project-dir
[--seuil-coche threshold] [--tol-marque tol]
[--list-fichiers files-list.txt | scan-files...]
π¦ θ§£θͺ¬ (DESCRIPTION) π¦
okay so the auto-multiple-choice analyse command does automatic mark recognition on scanned images of AMC multiple-choice exam answer sheets. Iconic. But listen β you cannot just roll up in here and run it cold π₯Ά. Before you even think about calling auto-multiple-choice analyse, you gotta:
- call
auto-multiple-choice-prepare(1)first to create the work documents (--mode s) and the MEP files (--mode m), and - process the layout with
auto-multiple-choice-meptex(1).
That's the pipeline, respect the pipeline π
π the directory options
--data data-dirβ tells it where the data files live (peepauto-multiple-choice-meptex(1)and friends for context). Default isproject-dir/data, easy.--cr project-cr-dirβ sets the directory where the images used for mark recognition get created. For the record: the zoomed box images land in thezoomssubdirectory, the name-field images are stored asname-*.jpgfiles, and the layout result images are stored aspage-*.jpgfiles. Very organized, we love a structured queen π--projet project-dirβ if you're not using--cr, then passing this option setsproject-cr-dirtoproject-dir/crfor you. Convenient!
π how to tell it which scan files to process (two ways, pick one)
- drop the filenames straight in as arguments, or
- list the filenames in a plain file (one filename per line, fr fr) and point at that file with
--list-fichiers files-list.txt.
π the mark-recognition parameter options
--prop proportionβ sets the proportion of the area of each box that gets used to decide whether there's a mark in it or not. Default value is0.8.--bw-threshold thresholdβ this one's used when converting grayscale scan images to black and white. Bigger value = more pixels get treated as black. Smaller value = more pixels get treated as white. The threshold must be between0and1(not optional, bestie π¨). Default value is0.6.--ignore-redβ with this option, only the red channel of a color scanned image is used. What that does: anything on the scan that was filled in with red gets completely ignored π. Super useful when the boxes themselves are printed in red.--tol-marque tolβ defines the tolerance when detecting the markers in the four corners of the scanned image. Detection works by hunting for black connected regions whose size is close to the target size target (the exact marker size you'd get if printing and scanning were absolutely perfect, which, lol, mood). Iftolis a plain real number, a size counts as "close" if it's between(1-tol)*targetand(1+tol)*target. Iftolis in the"tinf,tsup"form (wheretinfandtsupare real numbers), then the range is(1-tinf)*targetto(1+tsup)*target. Standard value is0.2.--multipleβ use this one when the exam questions were copied, i.e. several candidates could be answering the same exam questions. In that case each answer sheet gets assigned a copy number so the sheets sharing the same exam question number can be told apart π―--pre-allocate copy_idβ this one goes together with--multiple. The copy number for each scan starts at copy_id and gets assigned in the same order as the scan images you passed as arguments.--try-three | --no-try-threeβ says whether it should still process when only three of the four corner markers can be recognized. Sometimes life gives you three corners π--debug file.logβ names the file where the debug info gets written.
π θθ (AUTHORS) π
- Bienvenue Alexis[FAMILY Given]
<paamc@passoire.fr>β main author β¨ - Sarkar Anirvan[FAMILY Given] β author and editor
- Kagotani Hiroto[FAMILY Given] β editor
- BrΓ©al FrΓ©dΓ©ric[FAMILY Given] β author and editor
- BΓ©rard Jean[FAMILY Given] β French version β English version translation
- Khaznadar Georges[FAMILY Given] β French version β English version translation
- Kagotani Hiroto[FAMILY Given] β English version β Japanese version translation
the multilingual teamwork?? we stan π«Ά
π§ θ£½δ½θδ½ (COPYRIGHT) π§
Copyright Β© 2008-2018 Alexis BienvenΓΌe
This document can be used under the terms of the GNU General Public License version 2 or later.
heads up bestie: the license line is the actual legal bit, so don't go remixing it β GNU GPL v2 or later, exactly like it says π
okay that's the whole page, you're basically an AMC mark-recognition expert now, no cap π mwah, go scan some sheets β this was Auto Multiple Choice 1.7.0, dated 2025-04-17, signing off from your favorite AUTO-MULTIPLE-CHOI(1) bestie πππ¦
π¦ where this page comes from
- arch rolling: auto-multiple-choice 1.7.0-7 Β· GPL-2.0-or-later
- nixpkgs unstable: auto-multiple-choice 1.7.0
- π girlified by deepseek-ai/DeepSeek-V4.1-Flash-1017 on : a language model's rewrite of the original, under its terms