The proceedings contain 58 papers. The special focus in this conference is on Big Data, machine Learning, and applications. The topics include: A Comparative Study of Loss Functions for Deep Neural Networks in Time Se...
ISBN:
(纸本)9789819934805
The proceedings contain 58 papers. The special focus in this conference is on Big Data, machine Learning, and applications. The topics include: A Comparative Study of Loss Functions for Deep Neural Networks in Time Series Analysis;learning Algorithm for Threshold Softmax Layer to Handle Unknown Class Problem;traffic Monitoring and violation Detection Using Deep Learning;conjugate Gradient Method for finding Optimal Parameters in Linear Regression;rugby Ball Detection, Tracking and Future Trajectory Prediction Algorithm;early Detection of Heart Disease Using Feature Selection and Classification Techniques;Gun Detection System for Surveillance Cameras Using HOG-Assisted KNN Classifier;Optimized Detection, Classification, and Tracking with YOLOv5, HSv Color Thresholding, and KCF Tracking;realtime Object Distance Measurement Using Stereo visionimageprocessing;COvID-19 Detection Using Chest X-ray images;Comparative Analysis of LDA Algorithm for Low Resource Indian Languages with Its Translated English Documents;text Style Transfer: A Comprehensive Study on Methodologies and Evaluation;classification of Hindustani Musical Ragas Using One-Dimensional Convolutional Neural Networks;w-Tree: A Concept Correlation Tree for Data Analysis and Annotations;crawl Smart: A Domain-Specific Crawler;evaluating the Effect of Leading Indicators in Customer Churn Prediction;classification of Skin Lesion Using imageprocessing and ResNet50;data Collection and Pre-processing for machine Learning-Based Student Dropout Prediction;Nested Named-Entity Recognition in Multilingual Code-Switched NLP;an Insight on Drone applications in Surveillance Domain;deep Learning-Based Semantic Segmentation of Blood Cells from Microscopic images;a Partitioned Task Offloading Approach for Privacy Preservation at Edge;Artificial Intelligence in Radiological COvID-19 Detection: A State-of-the-Art Review.
Diabetic retinopathy (DR) is the leading cause of visual impairment globally. It occurs due to long-term diabetes with fluctuating blood glucose levels. It has become a significant concern for people in the working ag...
详细信息
Diabetic retinopathy (DR) is the leading cause of visual impairment globally. It occurs due to long-term diabetes with fluctuating blood glucose levels. It has become a significant concern for people in the working age group as it can lead to vision loss in the future. Manual examination of fundus images is time-consuming and requires much effort and expertise to determine the severity of the retinopathy. To diagnose and evaluate the disease, deep learning -based technologies have been used, which analyze blood vessels, microaneurysms, exudates, macula, optic discs, and hemorrhages also used for initial detection and grading of DR. This study examines the fundamentals of diabetes, its prevalence, complications, and treatment strategies that use artificial intelligence methods such as machine learning (ML), deep learning (DL), and federated learning (FL). The research covers future studies, performance assessments, biomarkers, screening methods, and current datasets. various neural network designs, including recurrent neural networks (RNNs), generative adversarial networks (GANs), and applications of ML, DL, and FL in the processing of fundus images, such as convolutional neural networks (CNNs) and their variations, are thoroughly examined. The potential research methods, such as developing DL models and incorporating heterogeneous data sources, are also outlined. Finally, the challenges and future directions of this research are discussed.
The human face is considered the prime entity in recognizing a person's identity in our society. Henceforth, the importance of face recognition systems is growing higher for many applications. Facial recognition s...
详细信息
The human face is considered the prime entity in recognizing a person's identity in our society. Henceforth, the importance of face recognition systems is growing higher for many applications. Facial recognition systems are in huge demand, next to fingerprint-based systems. Face-biometric has a highly dominant role in various applications such as border surveillance, forensic investigations, crime detection, access management systems, information security, and many more. Facial recognition systems deliver highly meticulous results in every of these application domains. However, the face identity threats are evenly growing at the same rate and posing severe concerns on the use of face-biometrics. This paper significantly explores all types of face recognition techniques, their accountable challenges, and threats to face-biometric-based identity recognition. This survey paper proposes a novel taxonomy to represent potential face identity threats. These threats are described, considering their impact on the facial recognition system. State-of-the-art approaches available in the literature are discussed here to mitigate the impact of the identified threats. This paper provides a comparative analysis of countermeasure techniques focusing on their performance on different face datasets for each identified threat. This paper also highlights the characteristics of the benchmark face datasets representing unconstrained scenarios. In addition, we also discuss research gaps and future opportunities to tackle the facial identity threats for the information of researchers and readers.
processing of missing data by modern neural networks, such as CNNs, remains a fundamental, yet unsolved challenge, which naturally arises in many practical applications, like image inpainting or autonomous vehicles an...
详细信息
ISBN:
(纸本)9781665409155
processing of missing data by modern neural networks, such as CNNs, remains a fundamental, yet unsolved challenge, which naturally arises in many practical applications, like image inpainting or autonomous vehicles and robots. While imputation-based techniques are still one of the most popular solutions, they frequently introduce unreliable information to the data and do not take into account the uncertainty of estimation, which may be destructive for a machine learning model. In this paper, we present MisConv, a general mechanism, for adapting various CNN architectures to process incomplete images. By modeling the distribution of missing values by the Mixture of Factor Analyzers, we cover the spectrum of possible replacements and find an analytical formula for the expected value of convolution operator applied to the incomplete image. The whole framework is realized by matrix operations, which makes MisConv extremely efficient in practice. Experiments performed on various imageprocessing tasks demonstrate that MisConv achieves superior or comparable performance to the state-of-the-art methods.
Infrared and thermal images have been used widely in different security applications. One of the drawbacks of such images is low contrast and noisy images, which should be enhanced. We present a new image enhancement ...
详细信息
ISBN:
(纸本)9781510673854;9781510673847
Infrared and thermal images have been used widely in different security applications. One of the drawbacks of such images is low contrast and noisy images, which should be enhanced. We present a new image enhancement algorithm based on block-rooting processing with artificial multi-scale-exposure image fusion. The proposed block-based multi-scale enhancement method is based on a 3-D block-rooting transform domain technique comprised: finding similar blocks in the image by block-matching;block-grouping for different block sizes;applying 3-D block-matching image enhancement;decomposition of the weight map and multi-scale enhanced images into the Gaussian and Laplacian pyramids;fusion by multiplying multi-scale images and weights. A new stage is proposed to obtain a local-global estimate of high-contrast images, also used in the general artificial fusion model. Some presented experimental results illustrate the performance of the proposed method on the thermal image dataset compared with the traditional methods.
The integration of data storage and computing capabilities into a single physical component has led to the development of a microneuronal network system with high precision and speed rates. To achieve this system arch...
详细信息
The integration of data storage and computing capabilities into a single physical component has led to the development of a microneuronal network system with high precision and speed rates. To achieve this system architecture, bold innovations in the underlying hardware structure and neural network architecture are required. This article introduces an optoelectronic storage device based on the MoS2/h-BN/graphene van der Waals heterojunction transistor. At room temperature, the transistor exhibits an electron mobility of up to 340 cm(2)/(v center dot s) and a large storage window due to its unique van der Waals heterojunction. The transistor's reconfigurable nonvolatile optoelectronic properties enable the construction of logic gates, including "AND", "OR", "NAND", and "NOR". Leveraging these logic gates, a microneural network system is created that simulates future machinevisionapplications. The system achieves a remarkable recognition rate of 96.3% for images in a multidimensional color space, demonstrating the significant development potential of the microneural network system based on the MoS2/h-BN/graphene vdW heterojunction transistor in future machinevision.
Convolutional neural networks (CNNs) have become widely adopted for computer vision tasks. However, the vast amount of design choices and the complex interactions among their hyperparameters, which ultimately influenc...
详细信息
Automatic rice variety identification or quality analysis is a challenging task in imageprocessing and reflects advanced insights into agricultural research with the help of emerging computational technologies. It is...
详细信息
Automatic rice variety identification or quality analysis is a challenging task in imageprocessing and reflects advanced insights into agricultural research with the help of emerging computational technologies. It is the process of identifying the variety of the rice grains by matching them with the training dataset. It is an arduous task because the quality of rice grains is distinct from each other due to the availability of their numerous varieties in the market and unique inherent characteristics. Therefore, customers must identify the superior quality of rice from different available types in the market. This paper demonstrates an exhaustive and transparent perspective on the recent research studies for developing various identification systems using other techniques and a broad view towards this peculiar research area. The paper's main aim is to present in an organized way the related works on identification systems of rice and finally throws exposure on the synthesis analysis based on the research findings. This research study provides valuable and valuable assistance to novice researchers in the agricultural field by amalgamating the studies of various methods and techniques of feature extractions and classification required for automatic variety identification of rice. It is evident from the study that research work carried out on the automated variety identification systems with higher accuracy rates in deep learning using a conjunction of various features of rice is minimal as compared to other techniques and indeed presents a future direction.
Local C-v mapping is a method to analyze and visualize the dynamics of polarization switching in ferroelectric materials with nanoscale resolution. This method uses a probe electrode to measure the butterfly-shaped C-...
详细信息
ISBN:
(纸本)9798350371918;9798350371901
Local C-v mapping is a method to analyze and visualize the dynamics of polarization switching in ferroelectric materials with nanoscale resolution. This method uses a probe electrode to measure the butterfly-shaped C-v curve characteristic of ferroelectrics, and then repeats the measurement while scanning the probe to investigate the inplane distribution. This method enables the acquisition of a large amount of measurement data reflecting the spatial distribution of domain switching characteristics in a short time. On the other hand, we are still searching for a method to analyze the huge amount of data and extract meaningful information from it. So far, we have attempted to analyze the data using unsupervised cluster analysis to classify each pixel into a pre-specified number of clusters based on the similarity of the C-v curve shape. This time, we introduced a different imageprocessing method, more specifically, a differential filtering method, and attempted to extract information different from conventional methods.
Retail taxonomy classification provides hierarchical labelling of items and it has widespread applications, ranging from product on-boarding, product arrangement and faster retrieval. It is fundamental to both physica...
详细信息
ISBN:
(纸本)9798350370287;9798350370713
Retail taxonomy classification provides hierarchical labelling of items and it has widespread applications, ranging from product on-boarding, product arrangement and faster retrieval. It is fundamental to both physical space as well as e-commerce. Manual processing based on meta-data was adopted and more recently, image based approaches have emerged. Traditionally, hierarchical classification in retail domain is performed using feature extractors and using different classifier branches for different levels. There are two challenges with this approach: error propagation from previous levels which affects the decision-making of the model and the label inconsistency within levels creating unlikely taxonomy tree. Further, the training frameworks rely on large datasets for generalized performance. To address these challenges, we propose PMTL, a progressive multi-level training framework with logit-masking strategy for retail taxonomy classification. PMTL employs a level-wise training framework using cumulative global representation to enhance and generalize output at every level and minimize error propagation. Also, we have proposed logit masking strategy to mask all irrelevant logits of a level and enforce the model to train using only the relevant logits, thereby minimizing label inconsistency. Further, PMTL is a generalized framework that can be employed to any full-shot and few-shot learning scheme without bells and whistles. Our experiments with three datasets with varied complexity in full-shot and few-shot scenario demonstrates the effectiveness of our proposed method compared to the state-of-the-art.
暂无评论