Home |
Aim & Scope | Editorial Board | Call for Paper | Instruction for Authors |
Contact Us | Open Access | Indexing & Archiving |
Help & Support | Downloads | Paper Submission |
S.No. | Article Title & Authors (Volume 13, Issue 6, December - 2020) | Page Nos. | Status |
1. | Evaluation of the Number of Epochs in an Automated Covid-19 Detection System from X-Ray Images using Deep Transfer Learning Muhammad Shofi Fuad, Choirul Anam, Kusworo Adi, Muhammad Ardhi Khalif, and Geoff Dougherty International Journal of Advances in Engineering & Technology (IJAET), Volume 13 Issue 6, pp. 134-140, December 2020. ABSTRACT This study aims to evaluate the number of epochs for automated COVID-19 detection from X-ray images using GoogleNet architecture. We used 400 digital X-ray images, with 200 COVID-19 and 200 normal. Each digital image has 1989 x 1482 pixels. To match the pixel size of the GoogleNet architecture (224 x 224 x 3), the pixel size of the original images was resized accordingly. The 3 layers of images were obtained from different kernels. Augmentation techniques of reflection and translation were used to augment the data by maintaining the same labels in the learning process. The training was conducted several times with a learning rate of 0.0001. The training was carried out with different numbers of epochs, i.e. 18, 18, and 24, and the number of iterations was 32. 80% of the data were used for training, and 20% were used for testing. The main output measuring a parameter of the identification was "accuracy". The accuracy is 96.25% at epoch 18, 97.50% at epoch 24, and 100% at epoch 30. It is shown that increasing the epoch numbers will increase the prediction accuracy for COVID-19. |
134-140 | Online |
2. | Convolutional Neural Network based approach for Landmark Recognition Aditya Surana, Harshit Mathur International Journal of Advances in Engineering & Technology (IJAET), Volume 13 Issue 6, pp. 141-146, December 2020. ABSTRACT In recent years, the world has witnessed a tremendous increase in digital cameras and mobile devices which has led to an even greater increase in the number of photographs clicked every year on average. Tourists, among others, contribute a considerable share to this effect with 1.323 billion international tourist arrivals worldwide in 2017 alone. With this insurgence in the amount of photographs clicked, arises the problem of recognition. A person might wonder what that one landmark was that he once visited. Landmark recognition technology helps people better understand and organize their photographs by predicting a landmark label directly from image pixels. A possible solution for landmark recognition has been presented by application of deep learning techniques. More specifically, convolutional neural networks have been used for detection of these images, with the help of open source pre-trained models. A person submitting a snap of a landmark can get relevant information regarding the landmark according to the match found from the millions of images present in the database. |
141-146 | Online |