ABSTRACT
This paper proposes and implements a dedicated hardware accelerated real-time face-mask detection system using deep learning (DL). The proposed face-mask detection model (MaskDetect) was benchmarked on three embedded platforms: Raspberry PI 4B with either Google Coral USB TPU or Intel Neural Compute Stick 2 VPU, and NVIDIA Jetson Nano. The MaskDetect was independently quantised and optimised for each hardware accelerated implementation. An ablation study was carried out on the proposed model and its quantised implementations on the embedded hardware configurations above as a comparison to other popular transfer-learning models, such as VGG16, ResNet-50V2, and InceptionV3, which are compatible with these acceleration hardware platforms. The ablation study revealed that MaskDetect achieved excellent average face-mask detection performance with accuracy above 94% across all embedded platforms except for Coral, which achieved an average accuracy of nearly 90%. With respect to detection performance (accuracy), inference speed (frames per second (FPS)), and product cost, the ablation study revealed that implementation on Jetson Nano is the best choice for real-time face-mask detection. It achieved 94.2% detection accuracy and twice greater FPS when compared to its desktop hardware counterpart.
ABSTRACT
Measuring fluid characteristics is of high importance in various industries such as the polymer, petroleum, and petrochemical industries, etc. Flow regime classification and void fraction measurement are essential for predicting the performance of many systems. The efficiency of multiphase flow meters strongly depends on the flow parameters. In this study, MCNP (Monte Carlo N-Particle) code was employed to simulate annular, stratified, and homogeneous regimes. In this approach, two detectors (NaI) were utilized to detect the emitted photons from a cesium-137 source. The registered signals of both detectors were decomposed using a discrete wavelet transform (DWT). Following this, the low-frequency (approximation) and high-frequency (detail) components of the signals were calculated. Finally, various features of the approximation signals were extracted, using the average value, kurtosis, standard deviation (STD), and root mean square (RMS). The extracted features were thoroughly analyzed to find those features which could classify the flow regimes and be utilized as the inputs to a network for improving the efficiency of flow meters. Two different networks were implemented for flow regime classification and void fraction prediction. In the current study, using the wavelet transform and feature extraction approach, the considered flow regimes were classified correctly, and the void fraction percentages were calculated with a mean relative error (MRE) of 0.4%. Although the system presented in this study is proposed for measuring the characteristics of petroleum fluids, it can be easily used for other types of fluids such as polymeric fluids.
ABSTRACT
Fully connected representation learning (FCRL) is one of the widely used network structures in multimodel image classification frameworks. However, most FCRL-based structures, for instance, stacked autoencoder encode features and find the final cognition with separate building blocks, resulting in loosely connected feature representation. This article achieves a robust representation by considering a low-dimensional feature and the classifier model simultaneously. Thus, a new hierarchical subnetwork-based neural network (HSNN) is proposed in this article. The novelties of this framework are as follows: 1) it is an iterative learning process, instead of stacking separate blocks to obtain the discriminative encoding and the final classification results. In this sense, the optimal global features are generated; 2) it applies Moore-Penrose (MP) inverse-based batch-by-batch learning strategy to handle large-scale data sets, so that large data set, such as Place365 containing 1.8 million images, can be processed effectively. The experimental results on multiple domains with a varying number of training samples from â¼ 1 K to â¼ 2 M show that the proposed feature reinforcement framework achieves better generalization performance compared with most state-of-the-art FCRL methods.
ABSTRACT
Gradient descent optimization of learning has become a paradigm for training deep convolutional neural networks (DCNN). However, utilizing other learning strategies in the training process of the DCNN has rarely been explored by the deep learning (DL) community. This serves as the motivation to introduce a non-iterative learning strategy to retrain neurons at the top dense or fully connected (FC) layers of DCNN, resulting in, higher performance. The proposed method exploits the Moore-Penrose Inverse to pull back the current residual error to each FC layer, generating well-generalized features. Further, the weights of each FC layers are recomputed according to the Moore-Penrose Inverse. We evaluate the proposed approach on six most widely accepted object recognition benchmark datasets: Scene-15, CIFAR-10, CIFAR-100, SUN-397, Places365, and ImageNet. The experimental results show that the proposed method obtains improvements over 30 state-of-the-art methods. Interestingly, it also indicates that any DCNN with the proposed method can provide better performance than the same network with its original Backpropagation (BP)-based training.
