本研究針對藥品自動影像類別驗證功能進行開發。
本研究包含兩個關鍵技術:
(1) 任意軸向之物件偵測技術 : 本技術基於深度殘差網路結合特徵金字塔網路之物件偵測技術建構藥品偵測網路。
在輸入影像中找尋藥品之旋轉邊界框之位置，系統會將該位置進行計算獲得其旋轉角度，
接著將旋轉邊界框內的藥品影像進行軸向對齊後依據旋轉邊界框進行裁切後輸入至藥品匹配系統對藥品進行是否為預期類別之藥品影像驗證確認。
本技術之藥品偵測準確率達99.21\%在至少0.75的IOU之情況下,平均偵測所需時間為0.1秒，具快速且高精準度之表現；
(2) 藥品匹配技術 : 本技術基於全卷積特徵嵌入網路對偵測技術所偵測到之影像進行特徵嵌入後得到相對低維度之語意特徵，
並且結合任意軸向之物件偵測技術所得到之影像尺度特徵，進行相似度度量決定是否通過驗證。
本藥品匹配技術在21類藥品驗證任務上可達到平均0.047\%之錯誤接受率，平均所需花費之匹配時間為0.04秒。
本研究所用之資料庫為拍攝國內地方醫學中心所提供之藥品所建立之影像資料庫，其中包含21類藥品影像。
本研究可應用於醫學中心內需使用或配發藥品之部門，透過影像驗證功能，為醫學中心內部繁忙用藥過程提供一道安全之確認。

This research is aimed at the development of automatic drug image verification functions.
This research contains two key technologies:
(1) Arbitrary axis-aligned object detection technology :
This technology builds a drug detection network based on the object detection technology of deep residual network combined with feature pyramid network.
In the input image, find the coordinate of the rotating bounding box of the drug, the system will calculate the coordinate to obtain its rotation angle,
Then, the drug images in the rotating bounding box are axially aligned, cut according to the rotating bounding box,
and then input to the drug matching system to verify whether the drug is in the expected category.
The drug detection accuracy rate of this technology is 99.21\% in the case of at least 0.75 IOU, the average detection time is 0.1 seconds, with fast and high accuracy performance;
(2) Drug matching technology :
This technology is based on a fully convolution feature embedding network to embed the images detected by the detection technology to obtain semantic features of relatively low dimensions.
The image scale features that can be added by combining arbitrary axis-aligned object detection technology,
and the similarity measurement is performed to determine whether to pass the verification.
This drug matching technology can achieve FAR of 0.047\% on 21 types of drug verification tasks, and the average matching time required is 0.04 seconds.
The database used in this research is an image database created by shooting medicines provided by domestic local medical centers,
which contains 21 categories of drug images.
This research can be applied to the departments in the medical center that need to use or dispense drugs.
The image verification function provides a safe confirmation for the busy drug process inside the medical center.

[1] Latif A Kruer RM, Jarrell AS. Reducing medication errors in critical care: a multimodal approach. Clinical pharmacology : advances and applications, 6:117–126, 2014.
[2] David J Cullen, Bobbie Jean Sweitzer, David W Bates, Elisabeth Burdick, Amy Edmondson, and Lucian L Leape. Preventable adverse drug events in hospitalized patients: a comparative study of intensive care and general care units. Critical care medicine, 25(8):1289–1297, 1997.
[3] Insook Cho, Hyeok Park, Youn Jeong Choi, Mi Heui Hwang, and David W. Bates. Understanding the nature of medication errors in an icu with a computerized physician order entry system. PLOS ONE, 9(12):1–15, 12 2014.
[4] Saghaei M Ravaghi H Farzi S, Irajpour A. Causes of medication errors in intensive care units from the perspective of healthcare professionals. J Res Pharm Pract., 3:158–165, 2017.
[5] Li Chen Wen Li Hongbin Yang, Lin Xiang. Application of 2d code and pda in hospital drug management. Chinese Journal of Hospital Pharmacy, 35:2142–2145,2015.
[6] D.J. Liu Ying Li, X.S. Che. A study on the method of checking for label of medicine bottle. International Conference on Testing and Measurement, 2015.
[7] Rongjun Zhang, Yucheng He, Peng Zhang, Ying Hu, Yong Cao, and Jianwei Zhang. The bottle size detection system based on bayesian classifier. In 2015 IEEE International Conference on Information and Automation, pages 1157–1162. IEEE, 2015.
[8] Mingde Gong, Xianming Zhang, and Tao Ni. Cylindrical medicine bottle’s character recognition of automotive dispensing machine. In 2016 2nd International Conference on Artificial Intelligence and Industrial Engineering (AIIE 2016). At lantis Press, 2016.
[9] Hui Xu, Huifang Liu, and Dongjun Liu. A study on the automated checking method for curved label image of powder medicine bottle. In 2017 3rd International Conference on Control, Automation and Robotics (ICCAR), pages 410–413. IEEE, 2017.
[10] Soyeong Lee, Sunhae Jung, and Hyunjoo Song. Cnn-based drug recognition and braille embosser system for the blind. Journal of Computing Science and Engineering, 12(4):149–156, 2018.
[11] Hsien-Wei Ting, Sheng-Luen Chung, Chih-Fang Chen, Hsin-Yi Chiu, and Yow Wen Hsieh. A drug identification model developed using deep learning technologies: experience of a medical center in taiwan. BMC Health Services Research,20(1):1–9, 2020.
[12] Lei Liu, Zongxu Pan, and Bin Lei. Learning a rotation invariant detector with rotatable bounding box. arXiv preprint arXiv:1711.09405, 2017.
[13] Karen Simonyan and Andrew Zisserman. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014.
[14] Minghui Liao, Baoguang Shi, and Xiang Bai. Textboxes++: A single-shot oriented scene text detector. IEEE transactions on image processing, 27(8):3676–3690, 2018.
[15] Minghui Liao, Baoguang Shi, Xiang Bai, Xinggang Wang, and Wenyu Liu. Textboxes: A fast text detector with a single deep neural network. In Thirty First AAAI Conference on Artificial Intelligence, 2017.
[16] Wei Liu, Dragomir Anguelov, Dumitru Erhan, Christian Szegedy, Scott Reed, Cheng-Yang Fu, and Alexander C Berg. Ssd: Single shot multibox detector. In European conference on computer vision, pages 21–37. Springer, 2016.
[17] Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, and Andrew Rabinovich. Going deeper with convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 1–9, 2015.
[18] Xinyu Zhou, Cong Yao, He Wen, Yuzhi Wang, Shuchang Zhou, Weiran He, and Jiajun Liang. East: an efficient and accurate scene text detector. In Proceedings of the IEEE conference on Computer Vision and Pattern Recognition, pages 5551–5560, 2017.
[19] Kye-Hyeon Kim, Sanghoon Hong, Byungseok Roh, Yeongjae Cheon, and Minje Park. Pvanet: Deep but lightweight neural networks for real-time object detection. arXiv preprint arXiv:1608.08021, 2016.
[20] Yingying Jiang, Xiangyu Zhu, Xiaobing Wang, Shuli Yang, Wei Li, Hua Wang, Pei Fu, and Zhenbo Luo. R2cnn: rotational region cnn for orientation robust scene text detection. arXiv preprint arXiv:1706.09579, 2017.
[21] Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. In Advances in neural information processing systems, pages 91–99, 2015.
[22] Jianqi Ma, Weiyuan Shao, Hao Ye, Li Wang, Hong Wang, Yingbin Zheng, and Xiangyang Xue. Arbitrary-oriented scene text detection via rotation proposals. IEEE Transactions on Multimedia, 20(11):3111–3122, 2018.
[23] Xue Yang, Jirui Yang, Junchi Yan, Yue Zhang, Tengfei Zhang, Zhi Guo, Xian Sun, and Kun Fu. Scrdet: Towards more robust detection for small, cluttered and rotated objects. In Proceedings of the IEEE International Conference on Computer Vision, pages 8232–8241, 2019.
[24] Karl Pearson. Principal components analysis. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 6(2):559, 1901.
[25] Geoffrey E Hinton and Ruslan R Salakhutdinov. Reducing the dimensionality of data with neural networks. science, 313(5786):504–507, 2006.
[26] Jane Bromley, Isabelle Guyon, Yann LeCun, Eduard Sa¨ckinger, and Roopak Shah. Signature verification using a” siamese” time delay neural network. In Advances in neural information processing systems, pages 737–744, 1994.
[27] Raia Hadsell, Sumit Chopra, and Yann LeCun. Dimensionality reduction by learning an invariant mapping. In 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’06), volume 2, pages 1735–1742. IEEE, 2006.
[28] Florian Schroff, Dmitry Kalenichenko, and James Philbin. Facenet: A unified embedding for face recognition and clustering. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 815–823, 2015.
[29] Sergey Zagoruyko and Nikos Komodakis. Learning to compare image patches via convolutional neural networks. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2015.
[30] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE transactions on pattern analysis and machine intelligence, 37(9):1904–1916, 2015.
[31] Oriol Vinyals, Charles Blundell, Timothy Lillicrap, Daan Wierstra, et al. Matching networks for one shot learning. In Advances in neural information processing systems, pages 3630–3638, 2016.
[32] Flood Sung, Yongxin Yang, Li Zhang, Tao Xiang, Philip HS Torr, and Timothy M Hospedales. Learning to compare: Relation network for few-shot learning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1199–1208, 2018.
[33] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016.
[34] Tsung-Yi Lin, Piotr Dolla´r, Ross Girshick, Kaiming He, Bharath Hariharan, and Serge Belongie. Feature pyramid networks for object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2117–2125, 2017.
[35] Ross Girshick. Fast r-cnn. In Proceedings of the IEEE international conference on computer vision, pages 1440–1448, 2015.
[36] Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Dolla´r. Focal loss for dense object detection. In Proceedings of the IEEE international conference on computer vision, pages 2980–2988, 2017.
[37] Jake Snell, Kevin Swersky, and Richard Zemel. Prototypical networks for few-shot learning. In Advances in neural information processing systems, pages 4077–4087, 2017.
[38] Oriol Vinyals, Charles Blundell, Timothy Lillicrap, Daan Wierstra, et al. Matching networks for one shot learning. In Advances in neural information processing systems, pages 3630–3638, 2016.
[39] Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009.
[40] Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dolla´r, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In European conference on computer vision, pages 740–755. Springer, 2014.
[41] Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014.