Session and scheduling information are listed below. Select a session from the list and press "Go" to view the abstracts for that session.
|
Session
40
: Machine Learning in Display Manufacturing |
Display Manufacturing
; Artificial Intelligence Including Machine Learning for Imaging
|
Wednesday, May 15 / 10:40 AM - 12:00 PM / San Jose Convention Center, LL20A
Chair:
Andriy Romanyuk, Glas Troesch AG, Buetzberg, Switzerland
Co-Chair:
Kazutaka Hayashi, AGC Inc., Tokyo, Japan
40.1 - An Auto Monitoring Method of Laser Beam Shape and Size by Employing an AI and Computer Vision Algorithm (10:40 AM - 11:00 AM)
-
Sang-Hoon Lim, Youngjin Oh, Kyung-Jin Yoo
Samsung Display Gihung City South Korea
-
An AI agent program with AI technology to determine OK, NG of laser beam and beam-size measurement technology based on image processing is developed and applied to a production line. This program enables automatic management and history management of beam shape and size.
40.2 - Novel Gamma Prediction Algorithm for AMOLED Displays Based on a Residual Network Model (11:00 AM - 11:20 AM)
-
Chaofan Xu, Xuesong Tian, Kening Zheng, Canghong Wang, Dongfang Yang, Jian Li, Yifei Wang, Qiang Xie, Shuai Hou, Kyoungwon Lee, Changman Kim, Fei Shang
Chengdu BOE Optoelectronic Technology Co., Ltd. Chengdu China
-
To ensure the uniformity of display image quality, gamma tuning is necessary. This is particularly time-consuming, resulting in lower production efficiency. Especially for multi-frequency projects, gamma tuning is required for each frequency individually, and the takt time (TT) is multiplied. For multi-frequency projects, the authors propose a novel gamma prediction algorithm. Based on an AI model to predict gamma, a gamma value of 120HZ is obtained by using the traditional algorithm, and the gamma value is fed into the AI model to predict the remaining frequency, which greatly saves gamma TT. Experiments show that gamma tuning can save TT by about 20% by using the predicted results as the initial value. Writing the predicted value directly to the register saves TT by about 70% and the optical test is almost to the relevant specifications.
40.3 - Waveform Analysis System for GAN-Based Anomaly Detection of Coater Pressure in Photolithography (11:20 AM - 11:40 AM)
-
Junkyun Lim, Bongik Jang, Hongyul Choi, Jongwoo Ham, Jinhee Lee
Samsung Display Gyeonggi-do South Korea
-
A waveform analysis system is designed and implemented and applied to coater equipment in photolithography. The system extracts waveforms from millisecond data measured during the processing time of one panel in the equipment, and diagnoses whether equipment processing is normal through GAN-based waveform pattern analysis.
40.4 - Improving Visibility Coherence Between Auto-Macro Inspection and Auto-Visual Inspection Using AI Image Translation (11:40 AM - 12:00 PM)
-
Jewoon Woo, Sugwoo Jeong, Seokhyun Yoon
Samsung Display Corp. Yong-In South Korea
-
The authors enhanced the mura visibility of automatic inspection system (auto macro) images through pre-processing, followed by translation into an image similar to auto visual inspection (AVI) images using Pix2pix GAN. The auto-macro images exhibited similarities with AVI images in terms of overall luminance, gray distribution, and mura visibility intensity.