簡介 Introductions

• 本模組結合傳統機器視覺及深度學習之優勢，以期解決現有金屬加工製品表面特性及光源不均造成自動檢測不穩定之問題，後續可搭配雲端通訊進行異常資訊回饋，可提早發現製程異常減少損耗，達到節能及提升產品品質之功效。
• This module combines the advantages of traditional machine vision and deep learning. This can solve the problem of unstable automatic inspection caused by uneven surface characteristics and light sources of existing metalworking products. It can also be used with cloud communication for abnormal information feedback, which can detect process abnormalities early and reduce loss to save energy and improve product quality.

特色與創新 Characters and Innovations

• 結合傳統機器視覺並輔以深度學習改善曲面之金屬易受表面反光影響而形成紋理變異誤判問題，可達成自動檢測。
• 可於機邊運算獨立進行邊緣運算，也可結合5G或雲端通訊達成加速運算與即時製程回饋之價值。
• 國內首創3D螺旋切齒齒輪檢測機，可檢測黑皮、撞傷與崩齒等瑕疵。
• Combine traditional machine vision with deep learning to improve the misjudgment of the texture variation of curved metal due to reflective influence. It can achieve automatic inspection.
• Not only can perform edge computing, but also can combine 5G or cloud communication to achieve accelerated computing and real-time process feedback.
• The first 3D spiral gear inspection machine in Taiwan. Can be used to detect defects such as black skin, bumps and gear valanche.

應用與效益 Applications and Benefits

• 金屬加工製品、水五金、齒輪等產業之外觀品質鑑別與回饋系統。
• Exterior quality identification and feedback system for metal processing products, plumbing hardware, gears, and other industries.

聯絡 contact

• 李韋辰 Wei-Chen Li
• E-mail: Jeecool@itri.org.tw
• TEL:886-3-5916670
• FAX:886-3-5915939

更多資訊

• 更多技術介紹：https://www.automan.tw/aboutus/annual-report.aspx
• YouTube技術影片：https://www.youtube.com/channel/UCU3037vDY8r0m_rJ_U7lHvQ

金屬製品外觀品質AI鑑別與回饋模組  AI Identification and Feedback Module for Appearance Quality of Metal Products

雲端加速運算與回饋示意圖 Schematic Diagram of Cloud Computing and Feedback

齒輪AI瑕疵檢測系統