06. Meslek Yüksekokulları

Permanent URI for this communityhttps://hdl.handle.net/20.500.13091/1628

Browse

Browsing 06. Meslek Yüksekokulları by Publisher "Ismail Saritas"

Filter results by typing the first few letters
Now showing 1 - 2 of 2
  • Thumbnail Image
    Article
    Citation - Scopus: 18
    Fire Detection in Images Using Framework Based on Image Processing, Motion Detection and Convolutional Neural Network
    (Ismail Saritas, 2021) Taşpınar, Yavuz Selim; Köklü, Murat; Altın, Mustafa
    Fire detection in images has been frequently used recently to detect fire at an early stage. These methods play an important role in reducing the loss of life and property. Fire is not only chemically complex, but also physically very complex. The shape and color of the flame varies according to the type of fuel in the fire. This has made fire detection a very challenging problem. Advanced image processing algorithms are also needed to accurately detect fire. To solve this problem, a three-stage fire framework was created in this study. In the first stage, the flame region was extracted from the images containing the fire region with the basic image processing algorithms. At this stage, reduce brightness, HSL, YCbCr, median and herbaceous filters are applied successively to the image. Since the flame image has a polygonal structure by nature, the number of edges of the flame region has been found. In the second stage, the mobility feature of the flame was utilized. For this purpose, the mobility of the flame was determined by comparing the video frames containing the fire image. The CNN method was used to detect the fire in the images. The CNN model was trained with the transfer learning method using the Inception V3, SequeezeNet, VGG16 and VGG19 trained models. As a result of the tests of the models, 98.8%, 97.0%, 97.3% and 96.8% classification success were obtained, respectively. With the proposed fire detection framework, it is thought that the damage caused by the fire can be reduced by early detection of the fire and timely intervention. © 2021, Ismail Saritas. All rights reserved.
  • Thumbnail Image
    Article
    Citation - Scopus: 18
    Object Recognition With Hybrid Deep Learning Methods and Testing on Embedded Systems
    (Ismail Saritas, 2020) Taşpınar, Yavuz Selim; Selek, Murat
    Object recognition applications can be made with deep neural networks. However, this process may require intensive processing load. For this purpose, hybrid object recognition algorithms that can be created for the recognition of an object in the image and the comparison of the working time of these algorithms on various embedded systems are emphasized. While Haar Cascade, Local Binary Pattern (LBP) and Histogram Oriented Gradients (HOG) algorithms are used for object detection, Convolutional Neural Network (CNN) and Deep Neural Network (DNN) algorithms are used for classification. As a result, six hybrid structures such as Haar Cascade+CNN, LBP+CNN, HOG+CNN and Haar Cascade+DNN, LBP+DNN, HOG+DNN are developed. In this study, these 6 hybrid algorithms were analyzed in terms of success percentage and time, then compared with each other. Microsoft COCO dataset was used to train and test all these hybrid algorithms. Object recognition success of CNN was 76.33%. Object recognition success of Haar Cascade+CNN, one of the hybrid methods we recommend, with a success rate of 78.6% is higher than CNN and other hybrid methods. LBP+CNN method recognized objects in 0.487 seconds which is faster than any other hybrid methods. In our study, Nvidia Jetson TX2, Asus TinkerBoard, Raspbbery Pi 3 B+ were used as embedded systems. As a result of these tests, Haar Cascade+CNN method on Nvidia Jetson TX2 was detected in 0.1303 seconds, LBP+DNN and Haar Cascade+DNN methods on Asus Tinker Board were detected in 0.2459 seconds, and HOG+DNN method on Raspberry Pi 3 B+ was detected in 0.7153 seconds.. © 2020, Ismail Saritas. All rights reserved.