Making computers "read"...

Raúl Roa

...through WhatsApp

Disclaimer

Who's this guy

Chief Technology Officer

WE MAKE COMPUTERS DO AMAZING THINGS...

PUT STUFF WHERE THEY BELONG

MAKE THEM UNDERSTAND

what's this talk about?

CATS... OF COURSE

SERIOUSLY...

IF WE TRIED TO, WE COULD ACTUALLY MAKE COMPUTERS READ
TODAY

BUT FIRST...

Let's address the elefant in the room

I HACKZ, I KNOWZ API'S

I KNOWZ BUZZ WORDS!

"ARTIFICIAL INTELLIGENCE"

"MACHINE LEARNING"

"COMPUTER VISION"

BUT WHY ARE THESE THINGS IN OUR FACE NOW?

3 DRIVING FORCES...

  • COMPUTING POWER
     
  • DATA AVAILABILITY
     
  • BETTER ALGORITHMS

SURE... BUT DO YOU KNOW HOW IT WORKS?

JUST ADD THE OCR... RIGHT?

OCR IS A PIPELINE, NOT SALT

CHALLENGES

  • DATA COLLECTION
     
  • DATA CURATION
     
  • DATA QUALITY
     
  • NO ONE SIZE FITS ALL
     
  • SPEED vs ACCURACY

THE PIPELINE

Tesseract's engine pipeline

NOT SO LONG AGO...

WE RELIED IN APROXIMATIONS

WHAT ABOUT HANDWRITTEN TEXT THEN?

BUT WAIT... THERE'S MORE

DEEP LEARNING IS HERE TO SAVE US ALL...

BUT... I'M HERE FOR WHATSAPP

Our pipeline

Our steps

  • Image pre-processing
     
  • Image classification
     
  • Ocr engine execution
     
  • Post-processing

Pre-processing

Classification

Segmentation

OCR Engine

Post-processing

  • Contextual spell checking
     
  • Word segmentation
     
  • Word classification
     
  • Dense vector storage for similarity checks

Output

LET'S SEE IT IN ACTION

HASTA LA VISTA BABY!

Making computers "read" through WhatsApp

By Raúl G. Roa Gómez

Making computers "read" through WhatsApp

  • 161