工業技術研究院

技術簡介

發展滾動學習布面風格特徵，協助新布種快速找到候選的驗布參數並能提供驗布自動化流程，以控制人工複判檢出圖片數。

Abstract

The camera parameters and image feature parameters, named style parameters, should be tuned based on products while fabric inspection. The proper style parameters brings high inspection performance, to reduce underkill and overkill. The relation between style parameter file and images are built with historical data. This tool compares current inspected image and historical image, and suggest a proper style parameter file, which is the corresponding style parameter file of the most similar historical image. While inspecting new product, the operator reduces tuning time with this tool.

技術規格

發展滾動學習布面風格特徵，協助新布種快速找到候選的驗布參數並能提供驗布自動化流程，以控制人工複判檢出圖片數。

Technical Specification

The camera parameters and image feature parameters, named style parameters, should be tuned based on products while fabric inspection. The proper style parameters brings high inspection performance, to reduce underkill and overkill. The relation between style parameter file and images are built with historical data. This tool compares current inspected image and historical image, and suggest a proper style parameter file, which is the corresponding style parameter file of the most similar historical image. While inspecting new product, the operator reduces tuning time with this tool.

技術特色

發展滾動學習布面風格特徵，協助新布種快速找到候選的驗布參數並能提供驗布自動化流程，以控制人工複判檢出圖片數。

應用範圍

新產品於AOI取像所需檢驗候選參數比對之場域

接受技術者具備基礎建議（設備）

本工具為雲端服務，透過瀏覽器即可操作

接受技術者具備基礎建議（專業）

1. 熟悉AOI取像作業流程 2. 具資料分析應用背景之資訊人員

聯絡資訊