Accuracy Of Image Recognition Research Articles

Redox-Active Azulene-based 2D Conjugated Covalent Organic Framework for Organic Memristors.

As a conjugated and unsymmetric building block composed of an electron-poor seven-membered sp2 carbon ring and an electron-rich five-membered carbon ring, azulene and its derivatives have been recognized as one of the most promising building blocks for novel electronic devices due to its intrinsic redox activity. By using 1,3,5-tris(4-aminophenyl)-benzene and azulene-1,3-dicarbaldehyde as the starting materials, an azulene(Azu)-based 2D conjugated covalent organic framework, COF-Azu, is prepared through liquid-liquid interface polymerization strategy for the first time. The as-fabricated Al/COF-Azu/indium tin oxide (ITO) mem-ristor shows typical non-volatile resistive switching performance due to the electric filed induced intramolecular charge transfer effect. Associated with the unique memristive performance, a simple convolutional neural network is built for image recognition. After 8 epochs of training, image recognition accuracy of 80% for a neutral network trained on a larger data set is achieved.

Redox‐Active Azulene‐based 2D Conjugated Covalent Organic Framework for Organic Memristors

AbstractAs a conjugated and unsymmetric building block composed of an electron‐poor seven‐membered sp2 carbon ring and an electron‐rich five‐membered carbon ring, azulene and its derivatives have been recognized as one of the most promising building blocks for novel electronic devices due to its intrinsic redox activity. By using 1,3,5‐tris(4‐aminophenyl)‐benzene and azulene‐1,3‐dicarbaldehyde as the starting materials, an azulene(Azu)‐based 2D conjugated covalent organic framework, COF‐Azu, is prepared through liquid‐liquid interface polymerization strategy for the first time. The as‐fabricated Al/COF‐Azu/indium tin oxide (ITO) memristor shows typical non‐volatile resistive switching performance due to the electric filed induced intramolecular charge transfer effect. Associated with the unique memristive performance, a simple convolutional neural network is built for image recognition. After 8 epochs of training, image recognition accuracy of 80 % for a neutral network trained on a larger data set is achieved.

Compressing Face Images Using Genetic and Gray Wolf Meta-heuristic Algorithms Based on Variable Bit Allocation

In image processing, compression plays an important role in monitoring, controlling, and securing the process. The spatial resolution is one of the most effective factors in improving the quality of an image; but, it increases the amount of storage memory required. Based on meta-heuristic algorithms, this article presents a compression model for face images with block division and variable bit allocation. Wavelet transform is used to reduce the dimensions of high spatial resolution face images. In order to identify important and similar areas of identical macroblocks, genetic algorithms and gray wolves are used. A bit rate allocation is calculated for each block to achieve the best recognition accuracy, average PSNR, and SSIM. The CIE and FEI databases have been used as case studies. The proposed method has been tested and compared with the accuracy of image recognition under uncompressed conditions and using the common SPIHT and JPEG coding methods. Recognition accuracy increased from 0.18% for 16×16 blocks to 1.97% for 32×32 blocks. Additionally, the gray wolf algorithm is much faster than the genetic algorithm in reaching the optimal answer. Depending on the application type of the problem, the genetic algorithm or the gray wolf may be preferred to achieve the maximum average PSNR or SSIM. At the bit rate of 0.9, the maximum average PSNR for the gray wolf algorithm is 34.92 and the maximum average SSIM for the genetic algorithm is 0.936. Simulation results indicate that the mentioned algorithms increase PSNR and SSIM by stabilizing or increasing recognition accuracy.

Fine-Grained Pornographic Image Recognition with Multi-Instance Learning

Image has become an essential medium for expressing meaning and disseminating information. Many images are uploaded to the Internet, among which some are pornographic, causing adverse effects on public psychological health. To create a clean and positive Internet environment, network enforcement agencies need an automatic and efficient pornographic image recognition tool. Previous studies on pornographic images mainly rely on convolutional neural networks (CNN). Because of CNN’s many parameters, they must rely on a large labeled training dataset, which takes work to build. To reduce the effect of the database on the recognition performance of pornographic images, many researchers view pornographic image recognition as a binary classification task. In actual application, when faced with pornographic images of various features, the performance and recognition accuracy of the network model often decrease. In addition, the pornographic content in images usually lies in several small-sized local regions, which are not a large proportion of the image. CNN, this kind of strong supervised learning method, usually cannot automatically focus on the pornographic area of the image, thus affecting the recognition accuracy of pornographic images. This paper established an image dataset with seven classes by crawling pornographic websites and Baidu Image Library. A weakly supervised pornographic image recognition method based on multiple instance learning (MIL) is proposed. The Squeeze and Extraction (SE) module is introduced in the feature extraction to strengthen the critical information and weaken the influence of non-key and useless information on the result of pornographic image recognition. To meet the requirements of the pooling layer operation in Multiple Instance Learning, we introduced the idea of an attention mechanism to weight and average instances. The experimental results show that the proposed method has better accuracy and F1 scores than other methods.

An Improved Vehicle Fine-Grain Identification Algorithm Based on Stochastic Weight Average

In the field of motor vehicle recognition, the use of neural network models has become the standard, and the tuning of hyperparameters and loss functions has been shown to be an effective way to improve the performance of these models. However, when using classical convolutional network architectures (e.g., ImageNet) and training them on motor vehicle images with random labels, the overparameterization problem can lead to suboptimal results and an increased risk of recognition failure. P. Ismailova et al. proposed a solution to this problem with the use of weight averaging, which resulted in the development of the simple and effective Stochastic Weight Averaging (SWA) optimizer. In this paper, we apply the SWA method to optimize the original recognition model and demonstrate significant improvements in accuracy through the use of different learning rate schemes with various traditional optimizers. We also identify suitable hyperparameter values to enhance the model's generalization abilities through several experiments, reducing the waste of resources in the motor vehicle recognition task and improving the recognition accuracy of fine-grained images in general, thus increasing the efficiency of related fields.

ASL Recognition by the Layered Learning Model Using Clustered Groups

American Sign Language (ASL) images can be used as a communication tool by determining numbers and letters using the shape of the fingers. Particularly, ASL can have an key role in communication for hearing-impaired persons and conveying information to other persons, because sign language is their only channel of expression. Representative ASL recognition methods primarily adopt images, sensors, and pose-based recognition techniques, and employ various gestures together with hand-shapes. This study briefly reviews these attempts at ASL recognition and provides an improved ASL classification model that attempts to develop a deep learning method with meta-layers. In the proposed model, the collected ASL images were clustered based on similarities in shape, and clustered group classification was first performed, followed by reclassification within the group. The experiments were conducted with various groups using different learning layers to improve the accuracy of individual image recognition. After selecting the optimized group, we proposed a meta-layered learning model with the highest recognition rate using a deep learning method of image processing. The proposed model exhibited an improved performance compared with the general classification model.

Information technology for recognition of handwritten ukrainian letters and numbers using synthetic data sets

The paper considers several variants of the architecture of convolutional neural networks for recognizing isolated handwritten digits and Ukrainian letters. The results of recognizing different images containing letters and numbers were compared on models with different architectures. Several variants of rather complex architectures of neural networks were considered. Research was conducted with VGG16 and VGG19 architectures, ResNet or ResNetV2, MobileNet or MobileNetV2, DenseNet. The possibility of learning convolutional neural networks with the help of a synthetic data set built on the basis of handwritten or cursive fonts is shown. The size of the training data set significantly affects the reliability of character recognition. The data sets used in the work correspond in volume to the well-known EMNIST Letters dataset. The lower limit of the sample size, which provides acceptable recognition accuracy, was about 1500 characters per class. Reducing the sample by reducing the number of symbols per class leads to a significant decrease in recognition accuracy (from 90-100% accuracy of recognizing elements of real inscriptions to 40-60% with a 4-fold reduction in the sample size). Thus, when using deep neural networks to recognize letters or numbers, the reliability of recognizing elements of real inscriptions depended primarily on the size of the training data set. The accuracy of recognition of the test data set after training all variants of the models was quite high - 97-98% and higher. However, the generation of training data sets of small size — 300-500 images per class - practically did not provide reliable recognition. In general, when comparing the achieved recognition accuracy of real images and the model training speed, the best performance was provided by the DenseNet or ResNetV2 family models. Experiments with changing the optimization algorithm compared to Adam did not give any improvement in the accuracy and reliability of recognizing real samples. Increasing the number of model training epochs beyond the specified one also did not change the results.

ANALYSIS OF NEURAL NETWORK ARCHITECTURES FOR RECOGNITION OF HANDWRITTEN NUMBERS

The accuracy of contemporary systems designed for the recognition of handwritten numbers exhibits significant variability, contingent upon the technologies selected during the development of a specific method. Neural networks have supplanted algorithms reliant on statistical data and datasets for this task, primarily due to their better performance. Several recent publications addressing the challenges of handwritten digit recognition employing diverse types of neural networks have been examined. This paper delves into the analysis of the architecture of two prominent types: dense neural networks and convolutional neural networks along with their respective training methods applicable to the recognition of handwritten digits from images. The study's findings reveal that Dense Neural Networks (DNN), while applicable to the task at hand, demonstrate diminished performance as the size of the input image increases. Conversely, Convolutional Neural Networks (CNN) prove more adept at image analysis, owing to their convolution and pooling operations, thus establishing their priority in this context. Four distinct types of learning applicable to neural networks were explored, namely Supervised Learning, Unsupervised Learning, Semi-supervised Learning, and Reinforcement Learning. The investigation concludes that Supervised Learning using backpropagation emerges as the most fitting method for training neural networks to address the identified problem. The paper further explores the stages and principles underpinning the operation of the backpropagation method. Building upon the derived results, there is a plan to develop an application leveraging the identified architecture. This application aims to empower users with the capability to efficiently extract digital information from images, irrespective of their resolution. The envisioned outcome is an application that attains high accuracy and processing speed in image recognition and extraction.

SAR-AD-BagNet: An Interpretable Model for SAR Image Recognition Based on Adversarial Defense

Although deep neural networks (DNNs) have achieved good results on some common datasets of synthetic aperture radar (SAR) automatic target recognition (ATR), DNNs are opaque and difficult to interpret, which limits its practical application. In view of this, many interpretable models have been proposed in recent years. However, the previous interpretable models only represent transparent network structures and cannot be called real “interpretable” models. We think that the interpretability of a model contains two meanings: on the one hand, the decision process of the model is transparent, and on the other hand, the decision-making basis of the model should be reasonable and conform to human cognition. For this reason, we propose a new SAR image recognition model based on adversarial defense, namely SAR-AD-BagNet. Not only does it have a transparent decision-making process, but it also has a more reasonable basis for decision-making. In addition, the model also has high SAR image recognition accuracy and strong adversarial robustness.

Classification of whole slide images of breast histopathology based on spatial correlation characteristics

目的 病理学检查是明确乳腺癌诊断及肿瘤类型的金标准。深度神经网络广泛应用于乳腺病理全切片的诊断工作并取得了明显进展，但是现有大多数工作只是将全切片切割成小图像块，对每个图像块进行单独处理，没有考虑它们之间的空间信息。为此，提出了一种融合空间相关性特征的乳腺组织病理全切片分类方法。方法 首先基于卷积神经网络对病理图像块进行预测，并提取每个图像块有代表性的深层特征，然后利用特征融合将图像块及其周围图像的特征进行聚合，以形成具有空间相关性的块描述符，最后将全切片图像中最可疑的块描述符传递给循环神经网络，以预测最终的全切片级别的分类。结果 本文构建了一个经过详细标注的乳腺病理全切片数据集，并在此数据集上进行良性/恶性二分类实验。在自建的数据集中与3种全切片分类方法进行了比较。结果表明，本文方法的分类精度达到96.3%，比未考虑空间相关性的方法高出了1.9%，与基于热力图特征和基于空间性和随机森林的方法相比，分类精度分别高出8.8%和1.3%。结论 本文提出的乳腺组织病理全切片识别方法将空间相关性特征融合和RNN分类集成到一个统一模型，有助于提高图像识别准确率，为病理图像诊断工作提供了高效的辅助诊断工具。;Objective Pathological examination can be as the“gold standard”to interpret breast cancer diagnosis and its tumor types. To improve the treatment effect of breast cancer，accurate interpretation is beneficial to clarify the type of the disease early and effectively. However，it is time-consuming to check the pathological whole slide images（WSIs）manually，and the diagnosis result is easily affected by personal experience. Due to computer vision-based convolutional neural networks（CNNs）are applied to the classification task of histopathological images computer-aided diagnostic（CAD）techniques for digital images of histopathology has been developing intensively. However，it is still challenged to split largesized WSIs into small patches and each patch is processed individually without the spatial information between them. Also， the information-involved of the patches learned is not utilized in the small size patch feature aggregation process. To resolve these problems，we develop a spatial correlation-based classification method for breast histopathology images，which can re-identify the WSIs classification problem through deep learning feature fusion and recurrent neural network（RNN）based classification. Method Our demonstration consists of four aspects：1）pre-processing operation of WSIs，2）CNN-based breast patch prediction，3）image patch feature fusion，and 4）RNN-based WSIs classification. First，it is focused on a preprocessing operation on whole slide images of breast pathology，using a sliding window to cut the WSIs into patches of a suitable size in terms of a CNN. Second，CNN-based patch prediction is used to predict the possibility of cancer in each patch and each patch is encoded as a fixed-length feature. To reduce unpredictable class of cancerous patch being classified as non-cancerous patch，the ResNet34 is used as the patch classification model and penalty factors is added in Focal loss. Third，the feature fusion of image blocks can be recognized as a feature fusion niche that takes the entire grid of patches as input and the spatial correlation of patches and their surrounding patches are as feature fusion of all patches within the grid，thus block descriptors are formed. Feature fusion methods are involved in，including 1）Weight，2）Max， 3）Avg，4）Norm3，and 5）WeightNorm3. Finally，the RNN-based WSIs classification is used to pass the block feature descriptors with high probability of cancer in each WSIs to the RNN，and the RNN model is used to learn the sequence relationship between the image block feature sequences. It can expand the processing field of view of the classification model to multiple Blocks and optimize the classification accuracy of breast pathology WSIs. Additionally，we design and develop an online recognition system for assistance. Result A detailed annotated whole slide images dataset of breast pathology is constructed，and benign/malignant dichotomous classification experiments are carried out on this dataset. The results are compared to three WSIs classification methods in the self-constructed dataset，and it shows that the classification accuracy of the method can be reached to 96. 3%，which is 1. 9% higher than non-spatial correlation method. The classification accuracy of each is 8. 8% and 1. 3% higher compared to heat map features-based methods and spatial correlation and random forest-related methods. Conclusion The proposed breast pathology recognition method can improve image recognition accuracy and provide an efficient diagnostic aid for pathology image diagnosis work，which integrates feature fusion and RNN classification into a synthesized model.

A study on finger vein recognition with few samples based on residual connected meta‐learning

AbstractA meta‐learning based method for finger vein recognition with few samples is proposed to overcome the problem of low recognition accuracy caused by the small number and variety of finger vein samples as well as fuzzy vein lines. The method is based on meta‐learning, incorporating multiscale features, and using the idea of residual networks to join meta‐learning to improve the recognition accuracy of finger vein images with few samples; to further improve its recognition ability, a differential map is constructed in the form of a differential between the finger vein image of singular value decomposition and finger vein image. We are the first to apply meta‐learning to the field of finger vein recognition, to our knowledge, and the experiments show that this approach is superior, with recognition accuracy of up to 99.13% for finger vein datasets with few‐shot samples.

Optical synaptic devices with ultra-low power consumption for neuromorphic computing

Brain-inspired neuromorphic computing, featured by parallel computing, is considered as one of the most energy-efficient and time-saving architectures for massive data computing. However, photonic synapse, one of the key components, is still suffering high power consumption, potentially limiting its applications in artificial neural system. In this study, we present a BP/CdS heterostructure-based artificial photonic synapse with ultra-low power consumption. The device shows remarkable negative light response with maximum responsivity up to 4.1 × 108 A W−1 at VD = 0.5 V and light power intensity of 0.16 μW cm−2 (1.78 × 108 A W−1 on average), which further enables artificial synaptic applications with average power consumption as low as 4.78 fJ for each training process, representing the lowest among the reported results. Finally, a fully-connected optoelectronic neural network (FONN) is simulated with maximum image recognition accuracy up to 94.1%. This study provides new concept towards the designing of energy-efficient artificial photonic synapse and shows great potential in high-performance neuromorphic vision systems.

Visual Sensing Technology for Digital Image-Oriented Public Art Design

With the development of computer software technology, computer image processing software has become increasingly significant in the public art design process. It can help with graphic design inventiveness and color presentation, as well as improve the aesthetics of public art design. This study aims to analyze the use of digital image synthesis and visual sensing techniques in public art design, as well as to use Dongba paintings, which reflects Chinese culture, as image fusion research objects. The Dongba painting’s line drawing is modeled and drawn, and an element library is created. The characteristic information of Dongba paintings is extracted using line integration convolution (LIC) algorithm. The background texture elements of the image are fused using image quilting texture’s improved synthesis algorithm. To improve the image effect of the linked Dongba paintings, a local color transfer algorithm based on color transfer rights is presented. The results show that the optimized synthesis algorithm of image quilting textures can achieve the same texture synthesis effect as the original algorithm. The method can meet the needs of real-world applications while also significantly improving the original technique’s performance. The synthesis time is reduced by 87.67%, and the image recognition accuracy of the local color transfer algorithm reaches 88.9%. Compared with other algorithms, the recognition accuracy is improved by 6.4%. The proposed digital image synthesis technique can create good images with different foreground and backdrop combinations and can provide a diverse visual expression. This contributes to public art’s visual production.

Unsupervised learning‐based recognition and extraction for intelligent automatic video retrieval

AbstractImage queries‐based automatic recognition is a crucial element of video retrieval. Many recognition systems were proposed with different illustrations in image‐based video retrieval domains. However, intelligent automatic video retrieval recognition (IAVRR) systems development and implementation usage on video retrieval are recurring only at long intervals. Because software and hardware schemes and their updating strategies are used to necessitate automatic recognition of video retrieval image queries, to manage such functions as automated recognition of the image queries in video retrieval with a reduced degree of system memory usage, spots detection, maintenance time, identification of exact duplicate videos, etc. Simultaneously, to solve such problems, it is necessary to use video retrieval and recognition methods and algorithms that allow recognising image queries with a high degree of accuracy in a sufficient time in the conditions of interference and distortion in images. This research determination demonstrates a modified R‐ratio with Viola‐Jones (Mr. VJ) framework with the relevance of developing techniques and algorithms for automatic recognition of image queries. The proposed methods and their relevant algorithms are used to retrieve the highest real videos and ensure video retrieval's mathematical operation in the comparable protected system's operation. The protected system controls over the localisation, normalisation, parsing, segmentation and recognition are accomplished. Hence, the proposed R‐ratio analysis showed high efficiency in experiments on automatic recognition of image queries. Its application made it possible to provide a reasonably short time and high image recognition accuracy in conditions of their different locations. When using the proposed R‐ratio analysis methods with Viola‐Jones, the recognition time is about < 1000 ms. The retrieval accuracy is from 90% to 100% at the various quality of the image pixels.

Research on denoising method of chinese ancient character image based on chinese character writing standard model

Ancient documents are historical evidence of cultural inheritance, and the damage brought by natural and human factors to ancient documents is inevitable, resulting in the collected images of ancient Chinese characters containing a large amount of noise, which seriously affects the accuracy of subsequent image recognition and thus creates a great obstacle to the digitization of ancient documents. To address the complexity of ancient text structure, this paper proposes a Chinese ancient text image denoising method based on the Chinese character writing standard model. The method firstly adds four additional local branches based on the global branching, and uses the supplementary character detail information to weaken the phenomenon of strokes adhering to noise due to the lack of local details; secondly, it introduces the simulation noise of ancient documents to simulate the real ancient character image morphology, which can be used for the adversarial training of this method. In the training process, the minimum absolute value deviation, smoothing loss, structural consistency loss and the refined loss function formed by the adversarial loss are used to iteratively optimize the parameters. Finally, experiments prove that the model in this paper can increase the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) of the image by at least 23.8% and 11.4%, and the user evaluation index (UV) has also reached more than 80%.

Research on Image Recognition Technology Based on Machine Learning

With the rapid development of artificial intelligence technology, the network image recognition technology of machine learning is intelligent and production and life in computer vision system technology in China have gradually achieved significant success. On this basis, the computer network image recognition system is constructed by machine learning with image features such as cross-verification, fitting, accuracy, feature selection, and dimensionality reduction, and the results show that the machine learning algorithm improves the stability and multi-domain application of computer image recognition through cross-verification and deviation of decision tree units. Recognition technologies such as feature vector extraction, edge information, and texture features of machine learning algorithms improve the accuracy of image recognition.

An FPGA-based memristor emulator for artificial neural network

Emulating memristor via FPGA offers flexibility for quick prototyping and design space exploration because memristor manufacture and integration are currently insufficient for large-scale application in neural networks. Although important, the creation of fast and lightweight FPGA memristor emulators is still very difficult. In this work, a reconfigurable memristor emulator that can be implemented in digital circuits is proposed. With less than 1% hardware utilization and a 258 MHz operating frequency, the suggested solution has been successfully synthesized and confirmed on a Xilinx ZYNQ-7000 FPGA. Additionally, this work builds a memristor crossbar array for extracting digital picture information and then performs the digital recognition task using a quantitative artificial neural network in order to assess the synaptic function of the suggested memristor model. The experimental results show that the recognition accuracy of MNIST dataset images by the memristor network circuit reaches 91.89%.

Klasifikasi Image Tumbuhan Obat Sirih dan Binahong Menggunakan Metode Convolutional Neural Network (CNN)

Medicinal plants are types of plants that are often used by the community because they have many benefits, such as to prevent or cure various diseases. Betel and binahong are medicinal plants that are widely used by Indonesian people. One part of the plant that is commonly used to classify plant species is the leaf. The CNN method is a very common method used for image classification because this method produces the most significant accuracy in image recognition. The study used 900 data images with a comparison of training, validation, and testing data, namely 7:2:1. Based on the test results using epochs of 15, 20, and 25 on betel leaf and binahong images, it shows the greatest accuracy of 95.67% for test data. So it can be concluded that the Deep Learning technique with CNN is able to classify images of betel and binahong leaves well.

Fine-grained Image Recognition Method using Discriminative Region-based Data Augmentation

In order to reduce the complexity of the network model and improve the accuracy of image recognition, a fine-grained image recognition method using discriminative region-based data augmentation is proposed. The method obtains the discriminative regions of the image through the attention mechanism, and then performs diversity data augmentation based on the discriminative regions, including region crop, region drop and region mix, and then uses the generated augmented samples to train the network model, the backbone network of the model is ResNet50. The proposed method is tested on 4 commonly used fine-grained image datasets CUB-200-2011, Stanford Cars, FGVC Aircraft and Stanford Dogs, and achieves high accuracy. The experimental results show that the proposed method can improve the localization ability and feature extraction ability of the model for discriminative regions, and it is more lightweight and easier to implement.

Label Noise Learning Method for Metallographic Image Recognition of Heat-Resistant Steel for Use in Pressure Equipment.

In metallographic examination, spherular pearlite gradation, an important step in a metallographic examination, is the main indicator used to assess the reliability of heat-resistant steel. Recognition of pearlite spheroidization via the manual way mainly depends on the subjective perceptions and experience of each inspector. Deep learning-based methods can eliminate the effects of the subjective factors that affect manual recognition. However, images with incorrect labels, known as noisy images, challenge successful application of image recognition of deep learning models to spherular pearlite gradation. A deep-learning-based label noise method for metallographic image recognition is thus proposed to solve this problem. We use a filtering process to pretreat the raw datasets and append a retraining process for deep learning models. The presented method was applied to image recognition for spherular pearlite gradation on a metallographic image dataset which contains 422 images. Meanwhile, three classic deep learning models were also used for image recognition, individually and coupled with the proposed method. Results showed that accuracy of image recognition by a deep learning model solely is lower than the one coupled with our method. Particularly, accuracy of ResNet18 was improved from 72.27% to 77.01%.