在成衣產業中裁縫工作評估一直扮演著相當重要的角色，裁縫工時的評估與預測已成為經營成衣業核心作業流程。然而影響裁縫工時的因素大體上應包含：裁縫長度、接縫型態、裁縫機器種類、做工難易度、機器轉數(RPM)、每吋針數、特殊做工、布料材質、特殊零件與其他特殊製程等因素。過去類神經網路技術曾經成功地應用在各種產業上，它具有訓練、學習與模擬等功能並且可做出有如人類思考模式的最佳化與正確的判斷能力。本研究企圖使用倒傳遞網路的方式為一家本地成衣製造廠預測與評估裁縫工時，首先對高度相關的一群裁縫參數來建構網路輸出是一項重要的研究活動，完成後進一步訓練與測試網路之成為未來裁縫工時之用。不過如何對成衣業整體做正確且適合的評估預測實屬不易，本研究僅就案例成衣業者提出類神經網路方法可行性之探討，並非對整體產業完成通用模式。最後本研究發現類神經網路方法適合該成衣製造廠不同類型裁縫製程與不同款式的工時評估與預測。

Work measurement always plays an essential role in the apparel industry. Sewing time prediction is one of core process to run the apparel business. The factors influencing the sewing time include sewing length, seam types, machine types, difficulty grades, RPM of machines, stitches per inch, specific workmanship, types of cloth, special parts, and other specific processes. Neural network has been successfully applied in many fields, and they also have the characters and functions of training, learning, and simulation to make optimal and accurate judgments by stimulating the human thought model. The main purpose of this study is to use back-propagation network to predict sewing time for a case of a local company. First of all, the collection of the parameters having high correlation with the output of the network is very important task. After appropriate training and testing, the network can be applied in future sewing time applications. Nevertheless, how to predict the sewing time of each apparel style accurately and adequately is not an easy task. In this research, we aim to solve the problem with neural network tools. Finally, we found that it is suitable for the variety of sewing process and different kinds of styles in the apparel industry.