Clinical keratoconus (KCN) detection is a challenging and time-consuming task. In the diagnosis process, ophthalmologists must revise demographic and clinical ophthalmic examinations. The latter include slit-lamb, corneal topographic maps, and Pentacam indices (PI). We propose an Ensemble of Deep Transfer Learning (EDTL) based on corneal topographic maps. We consider four pretrained networks, SqueezeNet (SqN), AlexNet (AN), ShuffleNet (SfN), and MobileNet-v2 (MN), and fine-tune them on a dataset of KCN and normal cases, each including four topographic maps. We also consider a PI classifier. Then, our EDTL method combines the output probabilities of each of the five classifiers to obtain a decision based on the fusion of probabilities. Individually, the classifier based on PI achieved 93.1% accuracy, whereas the deep classifiers reached classification accuracies over 90% only in isolated cases. Overall, the average accuracy of the deep networks over the four corneal maps ranged from 86% (SfN) to 89.9% (AN). The classifier ensemble increased the accuracy of the deep classifiers based on corneal maps to values ranging (92.2% to 93.1%) for SqN and (93.1% to 94.8%) for AN. Including in the ensemble-specific combinations of corneal maps’ classifiers and PI increased the accuracy to 98.3%. Moreover, visualization of first learner filters in the networks and Grad-CAMs confirmed that the networks had learned relevant clinical features. This study shows the potential of creating ensembles of deep classifiers fine-tuned with a transfer learning strategy as it resulted in an improved accuracy while showing learnable filters and Grad-CAMs that agree with clinical knowledge. This is a step further towards the potential clinical deployment of an improved computer-assisted diagnosis system for KCN detection to help ophthalmologists to confirm the clinical decision and to perform fast and accurate KCN treatment.
<span lang="EN-US">The need for robotics systems has become an urgent necessity in various fields, especially in video surveillance and live broadcasting systems. The main goal of this work is to design and implement a rover robotic monitoring system based on raspberry pi 4 model B to control this overall system and display a live video by using a webcam (USB camera) as well as using you only look once algorithm-version five (YOLOv5) to detect, recognize and display objects in real-time. This deep learning algorithm is highly accurate and fast and is implemented by Python, OpenCV, PyTorch codes and the Context Object Detection Task (COCO) 2020 dataset. This robot can move in all directions and in different places especially in
... Show MoreThe novel coronavirus 2019 (COVID-19) is a respiratory syndrome with similar traits to common pneumonia. This major pandemic has affected nations both socially and economically, disturbing everyday life and urging the scientific community to develop solutions for the diagnosis and prevention of COVID-19. Reverse transcriptase-polymerase chain reaction (RT–PCR) is the conventional approach used for detecting COVID-19. Nevertheless, the initial stage of the infection is less predictable in PCR tests, making early prediction challenging. A robust and alternative diagnostic method based on digital computerised technologies to support conventional methods would greatly help society. Therefore, this paper reviews recent research bas
... Show MoreHigh peak to average power ration (PAPR) in orthogonal frequency division multiplexing (OFDM) is an important problem, which increase the cost and complexity of high power amplifiers. One of the techniques used to reduce the PAPR in OFDM system is the tone reservation method (TR). In our work we propose a modified tone reservation method to decrease the PAPR with low complexity compared with the conventional TR method by process the high and low amplitudes at the same time. An image of size 128×128 is used as a source of data that transmitted using OFDM system. The proposed method decrease the PAPR by 2dB compared with conventional method with keeping the performance unchanged. The performance of the proposed method is tested with
... Show MoreSmishing is a cybercriminal attack targeting mobile Short Message Service (SMS) devices that contains a malicious link, phone number, or email. The attacker intends to use this message to steal the victim's sensitive information, such as passwords, bank account details, and credit cards. One method of combating smishing is to raise awareness and educate users about the various tactics used by SMS phishers. But even so, this method has been criticized for becoming inefficient because smishing tactics are continually evolving. A more promising anti-smishing method is to use machine learning. This paper introduces a number of machine learning algorithms that can be used for detecting smishing. Furthermore, the differences and simil
... Show MoreThe current study involves placing 135 boreholes drilled to a depth of 10 m below the existing ground level. Three standard penetration tests (SPT) are performed at depths of 1.5, 6, and 9.5 m for each borehole. To produce thematic maps with coordinates and depths for the bearing capacity variation of the soil, a numerical analysis was conducted using MATLAB software. Despite several-order interpolation polynomials being used to estimate the bearing capacity of soil, the first-order polynomial was the best among the other trials due to its simplicity and fast calculations. Additionally, the root mean squared error (RMSE) was almost the same for the all of the tried models. The results of the study can be summarized by the production
... Show MoreArtificial intelligence techniques are reaching us in several forms, some of which are useful but can be exploited in a way that harms us. One of these forms is called deepfakes. Deepfakes is used to completely modify video (or image) content to display something that was not in it originally. The danger of deepfake technology impact on society through the loss of confidence in everything is published. Therefore, in this paper, we focus on deepfakedetection technology from the view of two concepts which are deep learning and forensic tools. The purpose of this survey is to give the reader a deeper overview of i) the environment of deepfake creation and detection, ii) how deep learning and forensic tools contributed to the detection
... Show MoreA hand gesture recognition system provides a robust and innovative solution to nonverbal communication through human–computer interaction. Deep learning models have excellent potential for usage in recognition applications. To overcome related issues, most previous studies have proposed new model architectures or have fine-tuned pre-trained models. Furthermore, these studies relied on one standard dataset for both training and testing. Thus, the accuracy of these studies is reasonable. Unlike these works, the current study investigates two deep learning models with intermediate layers to recognize static hand gesture images. Both models were tested on different datasets, adjusted to suit the dataset, and then trained under different m
... Show MoreThe necessities of steganography methods for hiding secret message into images have been ascend. Thereby, this study is to generate a practical steganography procedure to hide text into image. This operation allows the user to provide the system with both text and cover image, and to find a resulting image that comprises the hidden text inside. The suggested technique is to hide a text inside the header formats of a digital image. Least Significant Bit (LSB) method to hide the message or text, in order to keep the features and characteristics of the original image are used. A new method is applied via using the whole image (header formats) to hide the image. From the experimental results, suggested technique that gives a higher embe
... Show More