114‐2: Computer Vision and Agentic AI – New Paradigm Shift for Display Applications and Manufacturing
SID Symposium Digest of Technical Papers · 2026-06 · 57 (1) · 1587-1590
DOI: 10.1002/sdtp.70001_429 ↗Kurz gefasst
Fachartikel in der Zeitschrift SID Symposium Digest of Technical Papers, online erschienen am 11.09.2026. Verfasst von Dahai YU und 5 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben.
Abstract (Originalsprache)
While computer vision designed to study and represent the human visual system has achieved significant milestones with deep learning technology in face recognition, defect detection, and image enhancement. There are still several limitations such as the long‐tail problem, scaling issues, transfer learning, and cross‐domain generalization hinder broad practical deployment. In display manufacturing applications, a major bottleneck is the heavy reliance on well‐structured data for supervised learning; high accuracy labeled images for defects are not only insufficient in quantity but also prohibitively expensive to collect. Consequently, state‐of‐the‐art systems often fail in ways that humans do not, there is an urgent need to build systems that produce more human‐aligned behavior. Since 2021, Large Language Models (LLMs) and Agentic AI have undergone remarkable advancements, driving transformative developments across diverse domains such as natural language understanding, multimodal generation, and human‐AI interaction. Capitalizing on this emerging multimodal paradigm, both industry and academia are increasingly exploring how general‐purpose foundation models can be adapted to domain‐specific challenges. By leveraging their exceptional generalization capabilities, deep semantic reasoning, and zero‐or few‐shot adaptability, researchers are developing more efficient and robust AI agents for real world applications. In this paper, we explore the implications of this new paradigm for the display industry. We propose Tri‐Agent framework oriented multimodal models to address long‐standing, high‐impact challenges in the display manufacturing industry, specifically, automated defect detection, automated action recognition for operator guidance, and high‐fidelity 3D content generation for design and simulation. These results demonstrate that composing AI agent with vision‐language model is both more efficient and more effective than existing methods, paving the way for practical and scalable tool‐augmented systems.
Zitieren
APA
YU, D., LIU, Q., MENG, D., WEI, H., SHE, B., & CHEN, G. (2026). 114‐2: Computer Vision and Agentic AI – New Paradigm Shift for Display Applications and Manufacturing. SID Symposium Digest of Technical Papers, 57(1), 1587-1590. https://doi.org/10.1002/sdtp.70001_429
Vancouver
YU D, LIU Q, MENG D, WEI H, SHE B, CHEN G. 114‐2: Computer Vision and Agentic AI – New Paradigm Shift for Display Applications and Manufacturing. SID Symposium Digest of Technical Papers. 2026;57(1):1587-1590. doi:10.1002/sdtp.70001_429
BibTeX
@article{YU2026,
author = {YU, Dahai and LIU, Qiang and MENG, Di and WEI, Haoyu and SHE, Bihai and CHEN, Gang},
title = {114‐2: Computer Vision and Agentic AI – New Paradigm Shift for Display Applications and Manufacturing},
journal = {SID Symposium Digest of Technical Papers},
year = {2026},
volume = {57},
number = {1},
pages = {1587--1590},
doi = {10.1002/sdtp.70001_429}
}
EUVIMED