本課程重要性分兩方面說明，分別是工業檢測及智慧報工。眾所皆知工業檢測在工業產品之品質管控非常重要，而工業檢測涵蓋面很廣，包括了產品外型檢驗及加工動作分析，這些應用除了要求速度也要求精確，這時智慧終端計算變得很重要，需要有強大的智慧推論計算加速。另外如何將AI訓練結果移植至SOC晶片並與感測周邊結合，是本課程目標之一。本次計畫使用聯盟C4模組之智慧終端裝置影像處理晶片設計與C7透過慣性感測器結合一AISOC晶片實現一套人體活動辨識模組，可以辨識基本的站、坐、跑步、走路、上下樓梯等活動資訊。場域應用為工業生產之動作辨識及生產數據之數位化及智慧化。分二階段教授課程模組內容，及其在智慧製造之應用。The importance of this course is explained in two aspects, namely industrial testing and smart reporting. It is well known that industrial testing is very important in the quality control of industrial products, and the industrial testing filtration surface is very wide, including product appearance inspection and processing operation analysis. These applications not only require speed but also require accuracy. At this time, the smart end calculation has become very good. Importantly, it requires strong wisdom to recommend calculation acceleration. In addition, how to transplant AI training results to SOC chips and combine them with the sensing surroundings is one of the goals of this course. This project uses the intelligent terminal device image processing chip design of the Alliance C4 module and the C7 through a sexually sensitive sensor to combine an AISOC chip to realize a set of human activity recognition modules, which can identify basic standing, sitting, running, walking, up and down Activities information such as stairs. Field applications are used as the identification of industrial production operations and the digitalization and intelligence of production data. The content of the course module is divided into two stages and its application in intelligent manufacturing.
https://github.com/FootprintAI/kubeflow-workshop
HTTPS://GitHub.com/footprint AI/Crying be flow-workshop
| 評分項目 Grading Method | 配分比例 Grading percentage | 說明 Description |
|---|---|---|
| 期中考期中考 Midterm exam |
30 | |
| 專題實作一專題實作一 Topic implementation one |
20 | C7模組之動作辨識 |
| 專題實作二專題實作二 Special Implementation Two |
20 | C4模組智慧終端裝置影像處理晶片設計，包括終端裝置嵌入式系統架構與鏡頭視訊介面介紹、 |
| 專題實作三專題實作三 Special Implementation Three |
30 | 生產儀錶板魚眼影像數據，使用影像處理FPGA魚眼鏡頭環景影像處理，將影像正規化，輔以OCR進行生產數據萃取。 |
