Yazar "Karadag, Batuhan" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • Küçük Resim Yok
    Öğe
    Comparison of Neural Style Transfer Performance of Deep Learning Models
    (Gazi Univ, 2021) Karadag, Batuhan; Ari, Ali; Karadag, Muge
    Neural style transfer is one of the most studied topics in both academic and industrial fields. Quality and performance enhancement are among the most targeted goals in the studies. In this study, the performance of different CNN models in neural style transfer was investigated. Deep features were obtained using VGG16, VGG19 and ResNet50 models. Thanks to these attributes, a new target image is created by taking the content of the content image and the style of the style image. Adam, RMSprop and SGD optimization algorithms are used. In neural transfer studies, the best visual performance was obtained from VGG19 network model by using SGD optimization algorithm. The fastest neural style transfer in terms of time was obtained using the SGD optimization algorithm in the ResNet50 convolutional neural network model.
  • Küçük Resim Yok
    Öğe
    Object Detection with YOLOv7 Model on Smart Mobile Devices
    (Gazi Univ, 2023) Karadag, Batuhan; Ari, Ali
    The YOLOv7 model, which is one of the current object detection algorithms based on deep learning, achieved an average accuracy of 51.2% in the Microsoft COCO dataset, proving that it is ahead of other object detection methods. YOLO has been a preferred model for object detection problems in the commercial field since it was first introduced, due to its speed , accuracy. Generally, high-capacity hardware is needed to run deep learning-based systems. In this study, it is aimed to detect objects in smart mobile devices without using a graphic processor unit by activating the YOLOv7 model on the server in order to be able to detect objects in smart mobile devices, which have become one of the important tools of trade today. With the study, the YOLOv7 object detection algorithm has been successfully run on mobile devices with iOS operating system. In this way, an image taken on mobile devices or already in the gallery after any image is transferred to the server, it is ensured that the objects in the image are detected effectively in terms of accuracy and speed.