Submit memes. Help read the archive.
Correct what the machine missed. Add context the machine cannot know.
The archive gets better every time a human looks at it.
Found a good one? Drop it in.
Memes, screenshots, WhatsApp forwards, scans, Instagram posts.
Everything is optional except the file.
Reviewed before entering the archive. Want something removed? Email us.
The scanner is reading it now.
If it needs a human reading, it'll appear in Needs Human Eyes.
ID: —
These archive items need a human reading.
The machine got stuck. OCR failed. Language unclear. Region uncertain. Context unknown.
If you can read Urdu Nastaliq, Persian, Arabic, or just know what a meme means — help here.
Objects that human readers have improved.
Machine read one thing. A human corrected it. The archive learned.
This is not data labeling. This is not crowdsourcing work.
Khajistan preserves Middle Eastern meme culture before platforms erase it.
OCR fails on Urdu Nastaliq. Machines do not know what a Pakistani insult means.
They do not understand a shrine poster's subtext.
Humans do. You do.
Every correction you submit is stored as a training pair.
When enough clean pairs accumulate for a given script, the archive can be re-read with better accuracy.
We do not claim to train a model right now. We collect clean corrections.
That is how OCR improvements begin.
Submitted at least one transcription or translation.
Explained the cultural context behind a meme — something no machine can know.
Successfully corrected OCR in a non-Latin script.
Identified the meme format, origin, or moment. Placed it in time and place.
Multiple corrections accepted. Trusted reviewer. Your corrections can be promoted without waiting for a second review.