此研究探索了在iOS設備上設計水果辨識應用程式，該應用程式基於在Matlab計算環境中開發的深度卷積神經網路，並且手動轉換為Python上的Keras模型其結構與Matlab上的網路相同。使用的資料庫為不同的開放資料庫組合，包括Fruits-360、水果數據庫、FIDS30以及ALOI，目的是為了能夠開發能識別台灣大部分水果的iOS應用程序。由於Matlab和Xcode之間並不相容，我們將在Matlab計算環境中開發的深度卷積神經網路的訓練資訊導出到csv文件，進一步由Python代碼讀取，用於在此環境中建構相同架構的Keras模型，並最終轉製成CoreML，放進Xcode環境裡頭進而開發iOS App。我們最終使用的資料庫包括2019年6月更新的最新版本Fruits-360、對Fruits-360進行同類型水果彙整、選自ALOI資料庫的水果圖像。我們在Matlab中訓練了我們的數據。第一個的驗證準確率為99.91％，第二個的驗證準確度為99.99％，最後一個的驗證準確度為99.96％。

This work explores designing fruit recognition application on iOS devices based on deep convolutional neural networks developed in the Matlab computing environment and manually translated to equivalent models of python Keras. The integrated fruit dataset is a combination of different audited open datasets, including Fruit360, fruits-database, FIDS30, and ALOI for the purpose of developing an iOS App that is able to recognize most fruits in Taiwan. Due to incompatibility between Matlab and Xcode, adaptable interconnections in a convolutional neural network developed in Matlab are exported to cvs files, further read by python codes for building an equivalent Keras model in this work, and eventually translated by coreML for iOS App design under Xcode. The final datasets we used include the latest version of fruits-360 updated at June 2019, the extended 51-label fruits dataset, where the same kind of fruits have sorted out into one label, and the fruits images chosen from ALOI. We trained our data in Matlab. The validation accuracy of the first one is 99.91%, the validation accuracy of the second one is 99.99%, and the validation accuracy of the last one is 99.96%.

1 Introduction 1
1.1 Motivation 1
1.2 Relevant Issues in Neural Network 1
2 Method 5
2.1 Deep Neural Network Designer in Matlab 5
2.2 Deep Convolutional Neural Networks 7
2.3 Keras in Python 11
2.4 CoreML in Xcode 11
3 Experiment 13
3.1 Dataset Description 13
3.2 Transfer CNNs from Matlab to Xcode 21
3.2.1 Check Parameters Transfer 21
3.2.2 Check The Connection between Matlab and Keras 29
3.2.3 Applied to Fruit Datasets 34
3.3 Develop an App in Xcode 46
4 Conclusion 49
References 51

Inkyu Sa, Zongyuan Ge, Feras Dayoub, Ben Upcroft, Tristan Perez, Chris McCool, Deep Fruits: A Fruit Detection System Using Deep Neural Networks Sensors, August (2016)
Suchet Bargoti, James P. Underwood, Image Segmentation for Fruit Detection and Yield Estimation in Apple Orchards Journal of Field Robotics, February (2017)
Yuzhen Lu, Renfu Lu, Detection of Surface and Subsurface Defects of Apples Using Structured-Illumination Reflectance Imaging with Machine Learning Algorithms Transactions of the ASABE, September (2018)
Yunong Tian, Guodong Yang, Zhe Wang, Hao Wang, En Li, Apple detection during different growth stages in orchards using the improved YOLO-V3 model Computers and Electronics in Agriculture, February (2019)
Madeleine Stein, Suchet Bargoti, James Underwood, Image Based Mango Fruit Detection, Localisation and Yield Estimation Using Multiple View Geometry Sensors, November (2016)
Alexander Wendel, James Underwood, Kerry Walsh, Maturity estimation of mangoes using hyperspectral imaging from a ground based mobile platform Computers and Electronics in Agriculture, December (2018)
Zhaodi Wang, Menghan Hu, Guangtao Zhai, Application of Deep Learning Architectures for Accurate and Rapid Detection of Internal Mechanical Damage of Blueberry Using Hyperspectral Transmittance Data Sensors, April (2018)
Amin Nasiri, Amin Taheri-Garavand, Yu-Dong Zhang, Image-based deep learning automated sorting of date fruit Postharvest Biology and Technology, July (2019)
Hulin Kuang, Cairong Liu, Leanne Lai Hang Chan, Hong Yan, Multi-class fruit detection based on image region selection and improved object proposals Neurocomputing, March (2018)
Yu-Dong Zhang, Zhengchao Dong, Xianqing Chen, Wenjuan Jia, Sidan Du, Khan Muhammad, Shui-Hua Wang, Image based fruit category classification by 13-layer deep convolutional neural network and data augmentation Multimedia Tools and Applications, September (2017)
Khurram Hameed, Douglas Chai, Alexander Rassau, A comprehensive review of fruit and vegetable classification techniques Image and Vision Computing, December (2018)
Horea Muresan, Mihai Oltean, Fruit recognition from images using deep learning Acta Universitatis Sapientiae, Informatica June (2018)
Y. Song, C.A. Glasbey, G.W. Horgan, G. Polder, J.A. Dieleman, G.W.A.M. van der Heijden, Automatic fruit recognition and counting from multiple images Biosystems Engineering, February (2014)
Juntao Xiong, Zhen Liu, Rui Lin, Rongbin Bu, Zhiliang He, Zhengang Yang, Cuixiao Liang, Green Grape Detection and Picking-Point Calculation in a Night-Time Natural Environment Using a Charge-Coupled Device(CCD) Vision Sensor with Artificial Illumination Sensors, March (2018)
Israr Hussain, Qianhua He, Zhuliang Chen, Automatic Fruit Recognition Based on DCNN for Commercial Source Trace System IJCSA Vol.8, No.2/3, June (2018)
Guoxiang Zeng, Fruit and Vegetables Classification System Using Image Saliency and Convolutional Neural Network IEEE 3rd Information Technology and Mechatronics Engineering Conference (ITOEC), Chongqing,pp.613-616 (2017)
Zaw Min Khaing, Ye Naung, Phyo Hylam Htut, Development of Control System for Fruit Classification Based on Convolutional Neural Network IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering (EIConRus), Moscow and St. Petersburg, Russia, pp.1805-1807 (2018)
http://aloi.science.uva.nl
Chao-Yuan Tian, MatConvNet and Caffe Deep Learning of Pattern Recognition Apps on iOS Devices, January (2019)
Apple official, ”Converting Trained Models to Core ML”, Jan (2018)