近年來人工智慧發展迅速，許多企業紛紛投入機器學習的領域，尤其在影像辨識方面，其中船舶辨識在海上扮演一個很重要的角色，如海上交通管制、漁業管理、船舶搜救等等。海上交通事故中，船舶間的碰撞為事故發生裡比例最高的，而其中人為疏忽又占了大多部分，為了減少此類船舶碰撞事故的發生，本研究實作一個嵌入式船舶辨識系統，透過在伺服器上訓練多種機器學習模型，比較各種模型在嵌入式系統和電腦上的準確率和執行速度，選擇表現較好的模型實現在嵌入式系統上。此系統可簡易佈置在船舶上，當有其他船隻靠近時可警示操作人員，藉由此系統讓船舶在海上或入港時可降低碰撞事件的發生。

With the development of the artificial intelligence and machine learning, many companies have started to invest in machine learning, especially in image recognition. Nowadays, ship identification system plays a very important role at sea management, such as marine traffic control, fishery management, ship search and rescue etc. In marine traffic accidents, collisions between ships are the highest proportion of accidents and human negligence accounts for most of them. In order to reduce the occurrence of ship collision accidents, this study implements an embedded ship identification system. We train a variety of machine learning models and compare the accuracy and execution time per image of various models on embedded systems and computer, selecting the model with better performance to implement on embedded systems. The system can be easily placed on the ship and warns the operator when other ships are approaching, whereby the system can reduce the occurrence of collision events when the ship is at sea or entering port.

[1] J. R. A. Morillas, I. C. García, U. Zölzer, "Ship detection based on SVM using color and texture features," IEEE International Conference on Intelligent Computer Communication and Processing, 2015.
[2] T. H. Tran, T. L. Le, "Vision based boat detection for maritime surveillance," IEEE Int. Conference on Electronics,Information, and Communications 2016.
[3] S. Suzuki, Y. Mitsukura, T. Furuya, "Ship detection based on spatio-temporal features," IEEE 10th France-Japan/ 8th Europe-Asia Congress on Mecatronics, 2014.
[4] Klaser, A., and Marszalek, M., "A Spatio-temporal Descriptor based on 3D-gradients", 19th British Machine Vision Conference, pp.995-1004, 2008.
[5] S. Ren, K. He, R. Girshick, "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, pp. 1137-1149, 2017.
[6] W. Liu, D. Anguelov, D. Erhan, C. Szegedy, and S. Reed. SSD: Single shot multibox detector. arXiv:1512.02325v2, 2015.
[7] R. Girshick, "Fast R-CNN," IEEE International Conference on Computer Vision, pp. 1440–1448 2015.
[8] K. Eldhuset, "An automatic ship and ship wake detection system for spaceborne sar images in coastal regions," IEEE Trans. Geosciences and Remote Sensing, vol. 34 no. 4, pp. 1010–1019, Jul. 1996.
[9] C. Zhu, H. Zhou, R. Wang, and J. Guo, "A novel hierarchical method of ship detection from spaceborne optical images based on shape and texture features," IEEE Trans. Geosciences and Remote Sensing, vol. 48, no. 9, pp. 3446–3456, 2010.
[10] R. Wijnhoven, K. van. Rens, Egbert G. T. Jaspers, Peter H. N. de With, "Online Learning for Ship Detection in Maritime Surveillance," In Thirty-first Symposium on Inf!irmation Theory, pp.73-80, 2010.
[11] A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, and H. Adam. Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv:1704.04861, 2017
[12] C. Szegedy, W. Liu, Y. Jia, "Going deeper with convolutions," IEEE Conference on Computer Vision and Pattern Recognition 2015