Yazar "Naeem Oleiwi Al-Mahdi, Israa" seçeneğine göre listele

Listeleniyor 1 - 1 / 1
  • Küçük Resim
    Öğe
    CNN googlenet and alexnet architecture deep learning for diabetic retinopathy image processing and classification
    (İstanbul Gelişim Üniversitesi Lisansüstü Eğitim Enstitüsü, 2023) Naeem Oleiwi Al-Mahdi, Israa
    The objective of this study is to analyze the performance of the two widely used convolutional neural network (CNN) architectures, namely AlexNet and GoogLeNet, in order to determine their relative merits and demerits with regard to the classification of photographic images. The purpose of this test is to evaluate how effectively they can recognize and prepare for a variety of various kinds of pictures. Experiments and analyses include a broad variety of subjects, such as the implementation of algorithms, various pre-processing methods, the continuation of algorithms, and various strategies for fine-tuning. Investigation is conducted into a variety of training period settings, and the findings and graphs obtained from these investigations are compared. The findings indicate that both AlexNet and GoogLeNet have the capability of being utilized in the process of photo classification. AlexNet already has a performance advantage over its rivals after only six rounds of training. In the beginning, GoogleNet was not as accurate as Caffe, but it quickly caught up by engaging in enormous training repetitions. In general, the findings of the study imply that selecting the appropriate CNN architecture should be driven more by necessity than by personal taste. While AlexNet offers high accuracy with fewer epochs, GoogLeNet shows potential for higher performance with further training. This research not only helps academics and practitioners make more educated judgments when selecting models for image classification tasks, but it also increases our understanding of CNN architectures.