BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A Modular OCR Solution for Logographic Scripts: From Labeling to R
 ecognition and User Interface Design - Peichao Qin - Faculty of Asian and 
 Middle Eastern Studies
DTSTART:20250515T120000Z
DTEND:20250515T130000Z
UID:TALK229690@talks.cam.ac.uk
CONTACT:Jack Atkinson
DESCRIPTION:In recent years\, optical character recognition (OCR) has beco
 me increasingly efficient in recognizing real-world image-related data\, p
 articularly in contexts involving phonetic writing systems such as Latin-b
 ased or modern alphabetic scripts\, where there are a manageable number of
  categories or sufficiently labeled training data. However\, for early log
 ographic systems whose characters derive from pictographic origins\, such 
 as Chinese oracle bone inscriptions\, Egyptian hieroglyphs\, Mesopotamian 
 cuneiforms\, and Mayan glyphs\, there exist usually thousands of character
 s and even more graphic variants. As such\, the relevant OCR systems often
  suffer from data inefficiency and class imbalance\, presenting challenges
  for models like ResNet and other CNN-based networks. To make matters wors
 e\, historians and palaeographers constantly disagree on issues regarding 
 character decipherment and classification\, further complicating the proce
 sses of data labeling and dataset compilation. This talk will use Chinese 
 oracle bone script as a case study to demonstrate how to efficiently addre
 ss these challenges primarily through four stages of work: 1). font creati
 on for ancient characters via image vectorization\; 2). text encoding and 
 labeling using external relational tables\; 3). ResNet-based model trainin
 g using synthetic data augmentation\; 4). Product deployment using modern 
 web architectures such as React and Vue.js. You will be able to find part 
 of these work on: https://oracular.azurewebsites.net/.
LOCATION:JJ Thomson Seminar Room\, Maxwell Centre\, and on Zoom
END:VEVENT
END:VCALENDAR