Evrişimsel Sinir Ağı Modellerinde Özellik Seçim Yöntemlerini Kullanarak Çiçek Görüntülerinin Sınıflandırılması

Goruntu isleme yontem ve tekniklerinin gun gectikce daha iyi sonuc vermesi, ekolojik dengenin duyarliligi acisindan onem arz etmektedir. Bu makale ekolojik dengenin temel unsuru olan cicek goruntulerinin siniflandirilmasini ele almaktadir. Son zamanlarda cicek goruntuleri uzerinde derin ogrenme yontemlerinin kullanimi artmistir. Bu calismada, cicek goruntulerinin siniflandirilmasi icin internette erisime acik olan veri seti kullanilmistir. Veri seti 4326 goruntuden olusmaktadir. Elde edilen veri kumesinde ozellik cikarimi icin derin ogrenme modellerinden evrisimsel sini agi (ESA) kullanilmistir. ESA mimarilerinden AlexNet, VGG-16 ve VGG-19 modelleri kullanilmistir. Uc modelinde ortak ozelligi 1000 ozellik veren tam baglantili katmana sahip olmalaridir. Cicek goruntulerinden elde edilen ozellikler destek vektor makineleri (DVM) ile siniflandirilmis ve elde edilen sonuclar karsilastirilmistir. Karsilastirma sonucunda en iyi siniflandirma performansini VGG-16 mimarisi ile saglanmistir. Elde edilen siniflandirma dogruluk orani %86,56’dir. Sonraki asamada ESA mimarilerinin son tam baglantili katmanindan elde edilen 1000 ozellik birlestirilerek 3000 ozellik seti olusturuldu. Ardindan, ozellik secim yontemlerinden; Maksimum Bilgi Katsayisi (MBK), Ridge regresyonu ve Ozyinelemeli Ozellik Eleme (OOE) yontemleri kullanilarak ozellik sayisi 300’e dusurulmustur. Ozellik secim yontemleri ile cikartilan en verimli ozellikler DVM yontemi ile yeniden siniflandirilmistir. Siniflandirma basari orani yaklasik %4,54 artarak %91,10 olmustur. Bu calisma ile cicek goruntulerinin siniflandirilmasinda ESA mimarileri ile birlikte ozellik secim yontemlerinin kullaniminin etkili oldugu gozlemlenmistir.

[1]  Henry Medeiros,et al.  Apple flower detection using deep convolutional networks , 2018, Comput. Ind..

[2]  Junhao Wen,et al.  Fundus Image Classification Using VGG-19 Architecture with PCA and SVD , 2018, Symmetry.

[3]  Aboul Ella Hassanien,et al.  Linear discriminant analysis: A detailed tutorial , 2017, AI Commun..

[4]  Serdar Solak,et al.  Görüntü işleme teknikleri ve kümeleme yöntemleri kullanılarak fındık meyvesinin tespit ve sınıflandırılması , 2018 .

[5]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[6]  Qingjin Peng,et al.  The Tabu_Genetic Algorithm: A Novel Method for Hyper-Parameter Optimization of Learning Algorithms , 2019, Electronics.

[7]  Abdulkadir Sengur,et al.  Efficient deep features selections and classification for flower species recognition , 2019, Measurement.

[8]  Francois P. S. Luus,et al.  Active Learning with TensorBoard Projector , 2019, ArXiv.

[9]  Sungyoung Lee,et al.  uEFS: An efficient and comprehensive ensemble-based feature selection methodology to select informative features , 2018, PloS one.

[10]  Guo-Feng Jiang,et al.  Hydraulic traits are more diverse in flowers than in leaves , 2018, bioRxiv.

[11]  Jose A. Ramirez-Hernandez,et al.  Control of a re-entrant line manufacturing model with a reinforcement learning approach , 2007, ICMLA 2007.

[12]  Kaizhu Huang,et al.  Reducing and Stretching Deep Convolutional Activation Features for Accurate Image Classification , 2018, Cognitive Computation.

[13]  Qiang Chen,et al.  Hierarchical matching with side information for image classification , 2012, 2012 IEEE Conference on Computer Vision and Pattern Recognition.

[14]  Alexander J. Smola,et al.  Efficient mini-batch training for stochastic optimization , 2014, KDD.

[15]  Heba Saadeh,et al.  Flower classification using deep convolutional neural networks , 2018, IET Comput. Vis..

[16]  M. Topal,et al.  Çoklu Doğrusal Bağlantı Durumunda Ridge ve Temel Bileşenler Regresyon Analiz Yöntemlerinin Kullanımı / Use of Ridge and Principal Component Regression Analysis Methods in Multicollinearity , 2013 .

[17]  Akif Dogantekin,et al.  A novel approach for liver image classification: PH-C-ELM , 2019, Measurement.

[18]  D. S. Guru,et al.  Textural features in flower classification , 2011, Math. Comput. Model..

[19]  Zafer Cömert,et al.  Fetal Hypoxia Detection Based on Deep Convolutional Neural Network with Transfer Learning Approach , 2018, CSOS.

[20]  Jeremy R. B. Newman,et al.  Variable selection in omics data: A practical evaluation of small sample sizes , 2018, PloS one.

[21]  C.-C. Jay Kuo.  Understanding convolutional neural networks with a mathematical model , 2016, J. Vis. Commun. Image Represent..

[22]  Yang Wang,et al.  Applications of Support Vector Machine (SVM) Learning in Cancer Genomics. , 2018, Cancer genomics & proteomics.

[23]  Luc Van Gool,et al.  TriCoS: A Tri-level Class-Discriminative Co-segmentation Method for Image Classification , 2012, ECCV.

[24]  Patrick Mäder,et al.  Plant species classification using flower images—A comparative study of local feature representations , 2017, PloS one.

[25]  Mostafa Mehdipour-Ghazi,et al.  Plant identification using deep neural networks via optimization of transfer learning parameters , 2017, Neurocomputing.

[26]  Joachim Denzler,et al.  Deep learning and process understanding for data-driven Earth system science , 2019, Nature.

[27]  M. Christenhusz,et al.  The number of known plants species in the world and its annual increase , 2016 .

[28]  Bu Sung Lee Francis,et al.  Combining MIC feature selection and feature-based MSPCA for network traffic anomaly detection , 2016, 2016 Third International Conference on Digital Information Processing, Data Mining, and Wireless Communications (DIPDMWC).

[29]  Yi Zhang,et al.  A Novel Algorithm for the Precise Calculation of the Maximal Information Coefficient , 2014, Scientific Reports.