China
Africa
Explore chapters and articles related to this topic
Deep Learning Approaches to Cancer Diagnosis using Histopathological Images
Published in Hassan Ugail, Deep Learning in Visual Computing, 2022
As explained earlier, we took a total number of 1,424 images were fed into the CNN model, by resizing the images according to the size of the image input layer. The AlexNet model requires input images of size 227 x 227 x 3, where 3 is the number of colour channels. VGG, GoogLeNet, and ResNet architectures require input images of size 224 x 224 x 3, while Incv3 and IncResNet models require input images of size 229 x 229 x 3. After that, the images were divided randomly into a training set of 80% (1,140 images) and a testing set of 20% (285 images), and then they were rotated (90, 180 and 270), flipped left to right horizontally, and then vertically. The small overlapping patches of sizes 32 x 32 were cropped along the images. Data augmentation usually helps to prevent the network from overfitting and to memorise the exact details of the training images.
Automatic Mask Detection and Social Distance Alerting Based on a Deep-Learning Computer Vision Algorithm
Published in S. Prabha, P. Karthikeyan, K. Kamalanand, N. Selvaganesan, Computational Modelling and Imaging for SARS-CoV-2 and COVID-19, 2021
N. Vinoth, A. Ganesh Ram, M. Vijayakarthick, S. Meyyappan
The architecture of the Convolutional Neural Network model consists of many convolution layers, pooling layers and fully-connected layers, as shown in Figure 5.1 (Huang et al., 2018). Filtering and smoothing operations are carried out in these layers of CNN. AlexNet is the basis of modern CNN, which is comprised of eight convolution layers, three pooling layers, and three fully connected layers, with Relu as the activation function. A plethora of improvisations, inspired by AlexNet, came into existence: VGGNet; GoogleNet; and ResNet. These accumulated convolution layers, pooling layers and fully connected layers (Szegedy et al., 2016). The purpose of generic object detection is to locate and identify current objects in any single image and to labeling them with rectangular bounding boxes to signify the confidences of existence (Uijlings et al., 2013).
Investigation of the Fingernail Plate for Biometric Authentication using Deep Neural Networks
Published in Gaurav Jaswal, Vivek Kanhangad, Raghavendra Ramachandra, AI and Deep Learning in Biometric Security, 2021
Surabhi Hom Choudhury, Amioy Kumar, Shahedul Haque Laskar
AlexNet [25] has been originally trained on a subset of the ImageNet database [26], which originally contained more than 15 million annotated images segregated into more than 22,000 categories. AlexNet consists of eight weighted layers; specifically five convolutional layers followed by three fully connected layers (fc6, fc7, fc8). The weighted layers are followed by one or more layers like Rectified Linear Units (ReLU) activation function, maxpooling function, Local Response Normalisation (LRN) function, etc. The output of the fc8 layer is provided to a softmax layer. This capacitates the network to predict what probability the test subject has, of belonging to the different trained classes. Due to the reasonably smaller size of the current database, building a new deep learning network would prove to be unproductive. Thus, Transfer Learning has been opted for.
Fault classification using deep learning based model and impact of dust accumulation on solar photovoltaic modules
Published in Energy Sources, Part A: Recovery, Utilization, and Environmental Effects, 2023
Rahma Aman, M. Rizwan, Astitva Kumar
AlexNet’s deep layers, comprised of 650 thousand neurons and 60 million parameters, allow it to categorize more than 1000 distinct classes. Alexnet consist of eight layers out of which five layers are convolutional layers (CLs) and three fully connected layers (FLCs). Input to the First convolution layer is 227 × 227 × 3 to the 1st convolutional layer have 96 filters of size 11 × 11 with stride of 4 pixel after the 1st convolutional layer pooling layer are connected with stride of 2 pixel this pooling layer is sued to reduce the image size then this pooling layer acts as input to the next convolutional layer here there are 256 filters of size 5 × 5 with stride 1 then it is connected to the next pooling layer which will further reduce the size, then all the remaining three conventional layers are connected without pooling layer connected in between the conventional layer then three fully connected layer are connected where the last layer which uses softmax activation function for output calculation which produces distribution over 1000 class label. The dimension of the 147-input image for the AlexNet needs to be 227 × 227 × 3, and the first CL converts the input image with 96 kernels sized at 11 × 11 × 3 with a stride of four pixels. This image serves as the input to the second layer, and the other details are given in Figure 6.
Defects identification in raw jute fibre using convolutional neural network models
Published in The Journal of The Textile Institute, 2023
T. Nageshkumar, Prateek Shrivastava, Biplab Saha, A. Subeesh, D. B. Shakyawar, Gunasindhu Sardar, Jayanta Mandal
AlexNet has 60 million parameters and 650,000 neurons, consists of 8 layers (in which 5 convolutional layers and 3 fully connected layers). The first two convolutional layers are followed by normalization and a max-pooling layer, the third and fourth are connected directly, and the fifth convolutional layer is followed by a max-pooling layer. The output goes into a series of two fully connected layers, in which the second fully connected layer feed into a softmax classifier. In order to prevent overfitting in the fully connected layers, the authors employed a regularization method called ‘dropout’ with a ratio of 0.5 (Maeda-Gutiérrez et al., 2020). Another feature of the AlexNet model is the use of Rectified Linear Unit (ReLU), which is applied to each of the first seven layers.
An intelligent belt wear fault diagnosis method based on deep learning
Published in International Journal of Coal Preparation and Utilization, 2023
Bingjun Wang, Dongyang Dou, Ning Shen
AlexNet (Alex, Sutskever and Hinton 2012) has an 8-layer network structure of which the first five layers are convolutional, and the last three layers are fully connected. Among them, the first, second, and fifth convolutional layers are followed by a pooling layer. The network structure is shown in Figure 2. AlexNet’ s convolutional layers all use ReLU as the activation function. The convolution kernel sizes in the first five layers are 11x11, 5x5, 3x3, 3x3, and 3x3, and the pooling layer adopts overlapping pooling. After the picture is input, it is processed by the convolutional layers for feature extraction, then ReLU activation is applied and pooling is performed, followed by normalization. After AlexNet’ s 5-layer calculation, the features in the images are abstracted to classify the feature information. Finally, the classification result is output through the fully connected layer.