Computer Vision Networks Research Articles

Real time distracted driver detection using Xception architecture and Raspberry Pi

Researchers are concentrating on developing technologies to identify and caution drivers against driving while distracted because it is a major cause of traffic accidents. According to the National Highway Traffic Safety Administrator's report, distracted driving is to blame for roughly one in every five car accidents.Our goal is to create an accurate and dependable method for identifying distracted drivers and alerting them to their lack of focus. We take inspiration from the success of convolutional neural networks in computer vision to do this. Our strategy entails putting in place a CNN-based system that can recognize when a driver is distracted as well as pinpoint the precise cause of their preoccupation. Real-time detection, however, necessitates three apparently mutually exclusive requirements for an optimal network: a small number of parameters, high accuracy, and fast speed.

Transformer-Based Generative Adversarial Networks in Computer Vision: A Comprehensive Survey

Transformer-Based Generative Adversarial Networks in Computer Vision: A Comprehensive Survey

Object-centric Learning with Capsule Networks: A Survey

Capsule networks emerged as a promising alternative to convolutional neural networks for learning object-centric representations. The idea is to explicitly model part-whole hierarchies by using groups of neurons called capsules to encode visual entities, then learn the relationships between these entities dynamically from data. However, a major hurdle for capsule network research has been the lack of a reliable point of reference for understanding their foundational ideas and motivations. This survey provides a comprehensive and critical overview of capsule networks, which aims to serve as a main point of reference going forward. To that end, we introduce the fundamental concepts and motivations behind capsule networks, such as equivariant inference . We then cover various technical advances in capsule routing algorithms as well as alternative geometric and generative formulations. We provide a detailed explanation of how capsule networks relate to the attention mechanism in Transformers and uncover non-trivial conceptual similarities between them in the context of object-centric representation learning. We also review the extensive applications of capsule networks in computer vision, video and motion, graph representation learning, natural language processing, medical imaging, and many others. To conclude, we provide an in-depth discussion highlighting promising directions for future work.

Understanding Wireless ChannelsThrough NeRF2

Despite Maxwell's formulation of the electromagnetic wave laws over a century and a half ago, accurately modeling the transmission of RF signals within electrically complex environments continues to be a formidable challenge. This complexity arises from intricate interactions of RF signals with various obstacles, such as reflections and diffractions. Inspired by the effective utilization of neural networks in computer vision for representing optical fields, this paper introduces NeRF2, a novel neural radio-frequency radiance field. NeRF2 conceptualizes a continuous, volumetric scene function to interpret RF signal propagation. Uniquely, NeRF2, once trained with limited signal measurements, can predict the characteristics and reception of a signal at an arbitrary point, given the transmitter's location. Functioning as a physical-layer neural network, NeRF2 amalgamates learned statistical models with physical ray tracing to create synthetic datasets, tailored for the training needs of application-layer artificial neural networks (ANNs). This integration, termed "turbo-learning," combines authentic and synthetic datasets, significantly augmenting ANNs' training efficacy. Experimental results illustrate that turbo-learning can substantially improve performance, with an estimated 50% enhancement. The efficacy of NeRF2 is further underscored in its applications in indoor localization and 5G MIMO systems.

A review of convolutional neural networks in computer vision

In computer vision, a series of exemplary advances have been made in several areas involving image classification, semantic segmentation, object detection, and image super-resolution reconstruction with the rapid development of deep convolutional neural network (CNN). The CNN has superior features for autonomous learning and expression, and feature extraction from original input data can be realized by means of training CNN models that match practical applications. Due to the rapid progress in deep learning technology, the structure of CNN is becoming more and more complex and diverse. Consequently, it gradually replaces the traditional machine learning methods. This paper presents an elementary understanding of CNN components and their functions, including input layers, convolution layers, pooling layers, activation functions, batch normalization, dropout, fully connected layers, and output layers. On this basis, this paper gives a comprehensive overview of the past and current research status of the applications of CNN models in computer vision fields, e.g., image classification, object detection, and video prediction. In addition, we summarize the challenges and solutions of the deep CNN, and future research directions are also discussed.

Tourism image classification based on convolutional neural network SqueezeNet--Taking Slender West Lake as an example.

Tourism image classification plays an important role in the study of clarifying the real perception of tourism resources by tourists, which cannot be studied in depth by human vision alone. The development of convolutional neural networks in computer vision brings new opportunities for tourism image classification research. In this study, SqueezeNet, a lightweight convolutional neural network, was selected and improved on the basis of the original model for 3740 Slender West Lake tourism image datasets. It is found that the validation accuracy of the model is up to 85.75%, and the size is only 2.64 MB, which is a good classification effect. This reduces the parameters while ensuring high accuracy classification of tourism images, providing a more scientific reference for the study of tourism images and pointing out a new direction for the development and planning of tourism resources.

Generative adversarial networks in computer vision: image synthesis and manipulation

Abstract In this paper, we first use the hidden variable loss to generate an adversarial network to optimize the efficiency of the original GAN operation. Secondly, the optimized GAN algorithm is used to train the encoder and decoder to form a new image-processing GAN by combining with the self-encoder and to standardize the operation process of image synthesis. On this basis, the performance of the self-encoder GAN is compared with the original GAN algorithm, and the quality of images generated by the self-encoder GAN and other ways is tested. The test results show that compared with the original GAN framework, the self-encoder GAN has a 50% improvement in operational efficiency in terms of root-mean-square error and an 80% reduction in synthesized global relative error. The self-coder GAN based on this algorithm is 22.5% higher than the traditional BiGAN framework and 2.5% higher than the OGAN framework in terms of FID quality criteria of the generated images and produces 30% less data capacity than BiGAN. The generated images have an average IS score of 3.435, which is superior to other base algorithms.

Types of discourse disseminated by food influencers: Trends on Instagram in France, Germany, Italy, Spain, and the United Kingdom

This research examines the discourse of nutrition influencers on Instagram within five European countries (France, Germany, Italy, Spain, and the United Kingdom) during 2021. The research aims to analyse the visual and textual content disseminated by these influencers and compare the predominant narratives among them. The corpus consists of 98 food influencers, and computer vision networks were employed to analyse visual discourse, linguistic analysis software for text analysis, and quantitative and qualitative methods to analyse hashtags. While promoting global trends (such as vegetable juices or bowls made with natural and unprocessed cereals, nuts, and fruits) to a lesser extent, influencers from the five countries highlight traditional food from their respective cultures in their visual discourse. The predominant discourse in all countries promotes the need for physical change through exercise and nutrition, sometimes fostering a diet culture. Sports and fitness are present in the discourse of influencers from these five countries, predominantly emphasising weight loss, muscle gain, and sports training with specific diets. Sports plays a significant role, with a preponderance of photographs depicting physical transformations and the promotion of food products and sports brands through inspiring images. Spain stands out in this regard as having the most images showing physical changes, especially in women. Although no evidence has been discovered to suggest that the sponsorship of unhealthy foods affects consumers’ preferences for them, the results show the widespread promotion of diet-related goods and foods. Notably, the real food movement was not emphasized in this extensive data analysis.

Automatized Detection of Periodontal Bone Loss on Periapical Radiographs by Vision Transformer Networks.

Several artificial intelligence-based models have been presented for the detection of periodontal bone loss (PBL), mostly using convolutional neural networks, which are the state of the art in deep learning. Given the emerging breakthrough of transformer networks in computer vision, we aimed to evaluate various models for automatized PBL detection. An image data set of 21,819 anonymized periapical radiographs from the upper/lower and anterior/posterior regions was assessed by calibrated dentists according to PBL. Five vision transformer networks (ViT-base/ViT-large from Google, BEiT-base/BEiT-large from Microsoft, DeiT-base from Facebook/Meta) were utilized and evaluated. Accuracy (ACC), sensitivity (SE), specificity (SP), positive/negative predictive value (PPV/NPV) and area under the ROC curve (AUC) were statistically determined. The overall diagnostic ACC and AUC values ranged from 83.4 to 85.2% and 0.899 to 0.918 for all evaluated transformer networks, respectively. Differences in diagnostic performance were evident for lower (ACC 94.1-96.7%; AUC 0.944-0.970) and upper anterior (86.7-90.2%; 0.948-0.958) and lower (85.6-87.2%; 0.913-0.937) and upper posterior teeth (78.1-81.0%; 0.851-0.875). In this study, only minor differences among the tested networks were detected for PBL detection. To increase the diagnostic performance and to support the clinical use of such networks, further optimisations with larger and manually annotated image data sets are needed.

Renyi entropy analysis of a deep convolutional representation for texture recognition

Despite the recent success of convolutional neural networks in computer vision in general, texture images still pose an important challenge to those models, especially when dealing with textures “in the wild”. In this context, deep learning models can benefit from the introduction of features long known to be useful for texture modeling. And that is the case of entropy, a measure of texture regularity that has played an important role in classical computer vision. Based on this observation, here we propose an alternative analysis over deep convolutional neural features based on entropy for texture representation and, particularly, texture classification. More precisely, we couple a module to the convolutional backbone that locally computes the Renyi entropy of the latent representation at multiple levels. The rationale for using Renyi entropy is essentially two-fold: (1) It connects the concept of entropy with multifractal theory, another well explored measure especially in domains of application where we seek some physical interpretation of the descriptors, e.g., in medicine, biology, and others; (2) It has an extra degree of freedom (α parameter) that can be fine-tuned. The main contribution of this study is the development of a strategy that can improve the performance of convolutional neural networks in texture recognition tasks, adding low computational cost. The effectiveness of our method is verified in texture classification of benchmark datasets, as well as in a practical task of plant species identification. Our method achieves a competitive accuracy of 84.5% in the classification of KTH-TIPS-2b and 80.9% in FMD, which are two challenging benchmark databases. Besides, an accuracy of 91.6% is obtained in the plant identification application, to our knowledge outperforming all the results previously published on this task. In both scenarios, the proposed descriptors outperform several approaches from the state-of-the-art, confirming the method potential as a rich and robust solution for texture analysis in general. The results also suggest that information-theoretical measures like entropy can be a reliable source of information to compose a precise and robust latent representation of texture images.

Attention-Mechanism-Based Models for Unconstrained Face Recognition with Mask Occlusion

Masks cover most areas of the face, resulting in a serious loss of facial identity information; thus, how to alleviate or eliminate the negative impact of occlusion is a significant problem in the field of unconstrained face recognition. Inspired by the successful application of attention mechanisms and capsule networks in computer vision, we propose ECA-Inception-Resnet-Caps, which is a novel framework based on Inception-Resnet-v1 for learning discriminative face features in unconstrained mask-wearing conditions. Firstly, Squeeze-and-Excitation (SE) modules and Efficient Channel Attention (ECA) modules are applied to Inception-Resnet-v1 to increase the attention on unoccluded face areas, which is used to eliminate the negative impact of occlusion during feature extraction. Secondly, the effects of the two attention mechanisms on the different modules in Inception-Resnet-v1 are compared and analyzed, which is the foundation for further constructing the ECA-Inception-Resnet-Caps framework. Finally, ECA-Inception-Resnet-Caps is obtained by improving Inception-Resnet-v1 with capsule modules, which is explored to increase the interpretability and generalization of the model after reducing the negative impact of occlusion. The experimental results demonstrate that both attention mechanisms and the capsule network can effectively enhance the performance of Inception-Resnet-v1 for face recognition in occlusion tasks, with the ECA-Inception-Resnet-Caps model being the most effective, achieving an accuracy of 94.32%, which is 1.42% better than the baseline model.

A Presentation of Structures and Applications of Convolutional Neural Networks

This paper reviewed the history of convolutional neural networks, why and how they developed, and what inspired the scientists to design them. To make CNNs simpler to understand, we wrote about their characteristics and structures while introducing the basic units of convolutional neural networks, including training and modeling parameters and how they would affect the confidence and efficiency of the whole process, different kinds of layers and how they work, multiple pooling methods and loss functions with their formulas. This paper also included applications of convolutional neural networks in computer vision and natural language processing while specifying and analyzing the technologies in use to clarify this introduction. Challenges and future research directions of the convolutional neural networks were pointed out to help refine this technique.

Convolutional Neural Networks in Computer Vision for Grain Crop Phenotyping: A Review

Computer vision (CV) combined with a deep convolutional neural network (CNN) has emerged as a reliable analytical method to effectively characterize and quantify high-throughput phenotyping of different grain crops, including rice, wheat, corn, and soybean. In addition to the ability to rapidly obtain information on plant organs and abiotic stresses, and the ability to segment crops from weeds, such techniques have been used to detect pests and plant diseases and to identify grain varieties. The development of corresponding imaging systems to assess the phenotypic parameters, yield, and quality of crop plants will increase the confidence of stakeholders in grain crop cultivation, thereby bringing technical and economic benefits to advanced agriculture. Therefore, this paper provides a comprehensive review of CNNs in computer vision for grain crop phenotyping. It is meaningful to provide a review as a roadmap for future research in such a thriving research area. The CNN models (e.g., VGG, YOLO, and Faster R-CNN) used CV tasks including image classification, object detection, semantic segmentation, and instance segmentation, and the main results of recent studies on crop phenotype detection are discussed and summarized. Additionally, the challenges and future trends of the phenotyping techniques in grain crops are presented.

Applications of graph convolutional networks in computer vision

Applications of graph convolutional networks in computer vision

Aging Residual Factorization Machines: A Multi-Layer Residual Network Based on Aging Mechanisms

With the rapid development of recommendation systems, models and algorithms supporting the core of recommendation systems have emerged one after another, and researchers have attempted to optimize them. However, the structure of these models is complex. Popular deep neural networks often achieve the highest utilization of data by increasing the number of hidden layers, ignoring the problems of exploding and vanishing gradients and even the entire degradation of the networks. However, researchers pay too much attention to algorithms and models and do not consider the dataset itself. Methods for processing data and finding possible connections between the data and models have become new explorable points. Cold start is also a problem that researchers have been trying to solve and optimize since the birth of the recommendation system. Recent studies also provide good ideas for solving cold start, but the problem is that researchers still do not focus on datasets. In order to fill the gap in the exploration and research of datasets, this paper takes the long tail distribution and cold start problems that are common in recommendation systems such as the starting point, combines the residual network in computer vision with deep learning, and proposes the aging mechanism of datasets. In this paper, a multi-layer residual network based on aging mechanisms called Aging Residual Factorization Machines (ARFM) is proposed. Parallel experiments with other model algorithms are carried out on three datasets of different sizes and categories. Experimental results show that ARFM achieve performance advantages under the premise of different recommendation tasks.

On quantization of convolutional neural networks for image restoration

Recently, commercial vision sensors hit the mobile market. To achieve that, computer vision networks had to be quantized. However, this topic was not studied well for Image Signal processor (ISP) challenging image restoration tasks, being crucially important for hardware implementation, as well as for deployment on hardware accelerators, e.g. Neural Processors Units (NPU). In this paper, we studied the effect of the quantization of deep learning network on image quality. We tried various quantization on raw RGBW image demosaicing. Experimental results show that 12 bit weight quantization can sustain image quality at the similar level with floating-point network. 10 bit quantized network shows slight degradation in objective image quality and mild visual artifacts. If network weight’s bit-depth can be significantly reduced for computer vision tasks, our finding shows that it is not true for raw image restoration tasks: at least 10 bit weights are required to provide sufficient image quality. However, we can save some memory on feature maps bit-depth. We can conclude that network bit depth is critical for raw image restoration.

Automatic Arabic Grammatical Error Correction based on Expectation-Maximization routing and target-bidirectional agreement

Automatic Grammar Error Correction (GEC) detects and corrects various types of syntax, spelling, and grammatical errors. Different approaches such as rule-based, Statistical Machine Translation (SMT), and Neural Machine Translation (NMT) have been proposed. Among these approaches, NMT based on seq2seq multi-head attention (Transformer) performs the best. The key shortcoming of GEC seq2seq models with multiple encoder-decoder layers is that only the top layer is exploited in the subsequent processes. In addition, due to the exposure bias problem during inference, some of the previous target words are deleted and replaced by other words generated by the model itself, which leads to unsatisfactory output. This paper proposed GEC model based on seq2seq Transformer for low-resource languages such as Arabic to address these issues. Initially, we proposed a noising method for constructing synthetic parallel data to overcome the bottleneck arising from the lack of corpus. Furthermore, motivated by the success of capsule networks in computer vision, we used the Expectation-Maximization routing algorithm to dynamically aggregate information across layers in Arabic GEC. Moreover, to conquer the exposure bias problem, we introduced a bidirectional regularization term using Kullback-Leibler divergence in the training objective to improve the agreement between Right-to-left and Left-to-right models. Experiments performed on two benchmarks QALB-2014 and QALB-2015 showed that our proposed model achieved the best F1 score compared to the existing Arabic GEC systems.

ResTS: Residual Deep interpretable architecture for plant disease detection

Recently many methods have been induced for plant disease detection by the influence of Deep Neural Networks in Computer Vision. However, the dearth of transparency in these types of research makes their acquisition in the real-world scenario less approving. We propose an architecture named ResTS (Residual Teacher/Student) that can be used as visualization and a classification technique for diagnosis of the plant disease. ResTS is a tertiary adaptation of formerly suggested Teacher/Student architecture. ResTS is grounded on a Convolutional Neural Network (CNN) structure that comprises two classifiers (ResTeacher and ResStudent) and a decoder. This architecture trains both the classifiers in a reciprocal mode and the conveyed representation between ResTeacher and ResStudent is used as a proxy to envision the dominant areas in the image for categorization. The experiments have shown that the proposed structure ResTS (F1 score: 0.991) has surpassed the Teacher/Student architecture (F1 score: 0.972) and can yield finer visualizations of symptoms of the disease. Novel ResTS architecture incorporates the residual connections in all the constituents and it executes batch normalization after each convolution operation which is dissimilar to the formerly proposed Teacher/Student architecture for plant disease diagnosis. Residual connections in ResTS help in preserving the gradients and circumvent the problem of vanishing or exploding gradients. In addition, batch normalization after each convolution operation aids in swift convergence and increased reliability. All test results are attained on the PlantVillage dataset comprising 54 306 images of 14 crop species.

IX-ResNet: fragmented multi-scale feature fusion for image classification

With the continuous in-depth study of convolutional neural network in computer vision, how to improve the performance of network structure has been the focus of current research. Recent works have shown that multi-scale feature concatenation, shortcut connection and grouping convolution can effectively train deeper networks and improve the accuracy and effectiveness of the network. In this paper, we present a novel feature transformation strategy of fragmented multi-scale feature fusion. Moreover, an efficient modularized image classification network, IX-ResNet, is proposed based on this new strategy. IX-ResNet consists of many large isomorphic modules stacked in the form of residual network while Each large module can be composed of many small heterogeneous modules. The performance of IX-ResNet is verified on cifar-10, cifar-100 and ImageNet-1 K datasets, which indicates that IX-ResNet model using fragmented multi-scale feature fusion strategy can further improve accuracy compare to the original grouping convolution network ResNeXt with the same or even lower parameters.

CVCMFF Net: Complex-Valued Convolutional and Multifeature Fusion Network for Building Semantic Segmentation of InSAR Images

Building segmentation of synthetic aperture radar (SAR) images is a challenging task that has not been solved well. High-resolution interferometric SAR (InSAR) images can provide delicate textures and interferometric phase images useful for building segmentation. However, current semantic segmentation networks in computer vision cannot be directly applied in InSAR building segmentation tasks to get good results because of the InSAR images’ particularity. In this article, we present a novel complex-valued convolutional and multifeature fusion network (CVCMFF Net) specifically for building semantic segmentation of InSAR images. This CVCMFF Net not only learns from the complex-valued SAR images but also considers multiscale and multichannel feature fusion. It can effectively segment the layover, shadow, and background on both the simulated InSAR building images and the real airborne InSAR images. The segmentation performance of CVCMFF Net is significantly improved compared with those of other state-of-the-art networks. By feature visualization, the feature extraction rule and feature fusion mechanism of the network are explored. We hope that the proposed network can be beneficial to InSAR phase filtering, phase unwrapping, and information extraction in urban areas.