在台灣，藥盒外部必須清楚標示藥品成分及注意事項，以保障用藥安全。然而，感冒藥的外包裝經常因損壞而導致辨識困難。為解決此問題，本研究提出利用影像辨識技術，通過多任務學習模型同時識別藥物類別和名稱。
本研究旨在開發一個基於多任務學習（Multi-Task Learning）的影像辨識模型，以提升藥物辨識的準確性和效率。實驗設計並訓練了三種模型：藥物類別辨識模型、藥物名稱辨識模型和多任務學習綜合辨識模型。實驗結果顯示，藥物類別辨識模型在辨識上具備極高的準確性和精確度。然而，藥物名稱辨識模型在準確率方面有待進一步提高。
應用多任務學習技術的綜合辨識模型展示了在同時辨識藥物類別和名稱方面的顯著優勢。其結果顯示，在多任務環境下，模型能夠更有效地共享資源和特徵，提升了藥名識別的整體表現。
綜上所述，本研究所開發的藥物辨識模型在藥物類別和名稱分類上均取得了較顯著的性能提升。單任務模型在特定任務上表現優異，而多任務模型則在綜合任務場景中展現了更大的潛力，尤其在提升藥名識別準確性上有明顯改進。未來的工作將進一步優化訓練資料和模型算法，以提升模型的泛化能力和實際應用效果，並為多樣的藥物辨識需求提供全面的技術支持。

In Taiwan, the outer packaging of medication boxes must clearly indicate the drug's ingredients and precautions to ensure safe usage. However, the packaging of cold medications often gets damaged, making identification difficult. To address this issue, this study proposes using image recognition technology through a multi-task learning model to simultaneously identify drug categories and names.
This research aims to develop an image recognition model based on Multi-Task Learning (MTL) to enhance the accuracy and efficiency of drug identification. Three models were designed and trained: a drug category identification model, a drug name identification model, and a comprehensive multi-task learning model. The experimental results show that the drug category identification model exhibits extremely high accuracy and precision in identification. However, the drug name identification model requires further improvement in accuracy.
The comprehensive identification model utilizing multi-task learning technology demonstrates significant advantages in simultaneously identifying drug categories and names. The results indicate that in a multi-task environment, the model can effectively share resources and features, enhancing the overall performance of drug name recognition.
In conclusion, the drug identification model developed in this study achieved significant performance improvements in the classification of drug categories and names. The single-task models performed excellently in specific tasks, while the multi-task model displayed greater potential in comprehensive task scenarios, especially in improving the accuracy of drug name recognition. Future work will further optimize the training data and model algorithms to enhance the model's generalization ability and practical application effect, providing comprehensive technical support for various drug identification needs.

論文審定書 i
誌謝 ii
摘要 iii
Abstract iv
目錄 v
圖次 vii
表次 viii
第一章 緒論 1
第一節 研究背景 1
第二節 研究動機 2
第三節 研究目的 3
第二章 文獻探討 4
第一節 分類模型 4
第二節 卷積神經網路 4
第三節 遷移學習 5
第四節 數據增強 6
第五節 多任務學習 7
第三章 研究方法與步驟 8
第一節 研究流程 8
第二節 研究方法 9
第三節 評估模型與標準 11
第四章 研究結果與分析 13
第一節 資料來源蒐集 13
第二節 資料清理 13
第三節 評估模型 14
第四節 解釋模型 15
第五章 研究結論與建議 19
第一節 研究結論 19
第二節 研究限制與建議 19
參考文獻 21

Caruana, R. (1997). Multitask Learning. Machine Learning, 28(1), 41–75. https://doi.org/10.1023/A:1007379606734
He, K., Zhang, X., Ren, S., & Sun, J. (2015). Deep Residual Learning for Image Recognition (arXiv:1512.03385). arXiv. https://doi.org/10.48550/arXiv.1512.03385
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017). ImageNet classification with deep convolutional neural networks. Communications of the ACM, 60(6), 84–90. https://doi.org/10.1145/3065386
Perez, L., & Wang, J. (2017). The Effectiveness of Data Augmentation in Image Classification using Deep Learning (arXiv:1712.04621). arXiv. https://doi.org/10.48550/arXiv.1712.04621
PRIDE政策研究指標資料庫. (n.d.). Retrieved June 12, 2024, from https://pride.stpi.narl.org.tw/index
Ribeiro, M. T., Singh, S., & Guestrin, C. (2016). “Why Should I Trust You?”: Explaining the Predictions of Any Classifier (arXiv:1602.04938). arXiv. https://doi.org/10.48550/arXiv.1602.04938
Schmidt, R. M. (2019). Recurrent Neural Networks (RNNs): A gentle Introduction and Overview (arXiv:1912.05911). arXiv. https://doi.org/10.48550/arXiv.1912.05911
Selvaraju, R. R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., & Batra, D. (2020). Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization. International Journal of Computer Vision, 128(2), 336–359. https://doi.org/10.1007/s11263-019-01228-7
Xie, S., Girshick, R., Dollár, P., Tu, Z., & He, K. (2017). Aggregated Residual Transformations for Deep Neural Networks (arXiv:1611.05431). arXiv. https://doi.org/10.48550/arXiv.1611.05431
Zhai, S., Cheng, Y., Lu, W., & Zhang, Z. (2016). Doubly Convolutional Neural Networks (arXiv:1610.09716). arXiv. https://doi.org/10.48550/arXiv.1610.09716
Zhang, H., Wu, C., Zhang, Z., Zhu, Y., Lin, H., Zhang, Z., Sun, Y., He, T., Mueller, J., Manmatha, R., Li, M., & Smola, A. (2020). ResNeSt: Split-Attention Networks (arXiv:2004.08955). arXiv. https://doi.org/10.48550/arXiv.2004.08955