應用電腦視覺技術於袋裝水泥倉儲管理系統自動化

Abstract

本論文旨在探討應用電腦視覺技術於袋裝水泥倉儲管理系統自動化的可行性。本研究主要採用點雲(Point Cloud)技術建立3D模型，並選擇適當的監視器架設位置，藉此監控並識別水泥品名，進行數量計算與進出貨追蹤，並將攝影機訊息自動傳送到資料庫中，以實現自動化管理。此外，本研究採用物體偵測(Object detection)技術、OCR光學字元辨識 (Optical Character Recongnition)技術、人工智慧(Artificial intelligence)等，透過標註現有的資料來進行訓練和學習，然後建立深度學習模型來進行預測，並由機器學習辨識監視器中水泥資訊，包含：進出貨、品牌、數量等資料來進行貨物管理，最後結合庫存管理系統，將貨物資訊自動寫入資料庫中，提供銷貨管理系統使用。經由實驗結果顯示，本研究所提出的方法能夠有效地辨識各種廠牌的水泥之高準確率，及高度誤差，並且能夠在不同的營造業和水泥經銷商等領域廣泛應用。本研究所提出的自動化管理方法，將為營建倉儲管理系統的現代化升級提供有力的支持。

The purpose of this paper is to explore the feasibility of applying computer vision technology to automate the management system of bagged cement storage. This study primarily utilizes point cloud technology to build a 3D model and selects appropriate positions for installing monitors to monitor and identify cement types, perform quantity calculations, and track incoming and outgoing shipments. The camera information is automatically transmitted to a database to achieve automated management. In addition, this research employs object detection technology, Optical Character Recognition (OCR), artificial intelligence, and other techniques. By annotating existing data, training and learning are conducted to establish deep learning models for prediction. Machine learning is used to recognize cement information captured by the monitors, including shipment records, brands, quantities, etc., for effective inventory management. Finally, by integrating with an inventory management system, the cement information is automatically written into the database, providing support for sales management systems. Experimental results demonstrate that the proposed method can accurately identify various cement brands with high precision and minimal error. It can be widely applied in different construction industries and cement distributors. The automation management approach proposed in this study provides strong support for the modernization of construction storage management systems.