Search results for: deep learning image model

Authors: Alona Faktor

Abstract:

In this work, we present a NN-based computational model that can perform attention shifts according to high-level instruction. The instruction specifies the type of attentional shift using explicit geometrical relation. The instruction also can be of cognitive nature, specifying more complex human-human interaction or human-object interaction, or object-object interaction. Applying this approach sequentially allows obtaining a structural description of an image. A novel data-set of interacting humans and objects is constructed using a computer graphics engine. Using this data, we perform systematic research of relational segmentation shifts.

Keywords: cognitive science, attentin, deep learning, generalization

Procedia PDF Downloads 176

Authors: Feng Yang

Abstract:

Machine learning (ML), especially deep learning (DL), has been extensively applied to many applications in recently years and gained great success in solving different problems, including scientific problems. However, conventional ML/DL methodologies are purely data-driven which have the limitations, such as need of ample amount of labelled training data, lack of consistency to physical principles, and lack of generalizability to new problems/domains. Recently, there is a growing consensus that ML models need to further take advantage of prior knowledge to deal with these limitations. Physics-informed machine learning, aiming at integration of physics/domain knowledge into ML, has been recognized as an emerging area of research, especially in the recent 2 to 3 years. In this work, physics-informed ML, specifically physics-informed neural network (NN), is employed and implemented to estimate the displacements at x, y, z directions in a solid mechanics problem that is controlled by equilibrium equations with boundary conditions. By incorporating the physics (i.e. the equilibrium equations) into the learning process of NN, it is showed that the NN can be trained very efficiently with a small set of labelled training data. Experiments with different settings of the NN model and the amount of labelled training data were conducted, and the results show that very high accuracy can be achieved in fulfilling the equilibrium equations as well as in predicting the displacements, e.g. in setting the overall displacement of 0.1, a root mean square error (RMSE) of 2.09 × 10−4 was achieved.

Keywords: deep learning, neural network, physics-informed machine learning, solid mechanics

Procedia PDF Downloads 126

Authors: Hyohun Kim, Dongwha Shin, Yeonseok Kim, Ji-Su Ahn, Kensuke Nakamura, Dongeun Choi, Byung-Woo Hong

Abstract:

Artefacts are commonly encountered in the imaging process of clinical computed tomography (CT) where the artefact refers to any systematic discrepancy between the reconstructed observation and the true attenuation coefficient of the object. It is known that CT images are inherently more prone to artefacts due to its image formation process where a large number of independent detectors are involved, and they are assumed to yield consistent measurements. There are a number of different artefact types including noise, beam hardening, scatter, pseudo-enhancement, motion, helical, ring, and metal artefacts, which cause serious difficulties in reading images. Thus, it is desired to remove nuisance factors from the degraded image leaving the fundamental intrinsic information that can provide better interpretation of the anatomical and pathological characteristics. However, it is considered as a difficult task due to the high dimensionality and variability of data to be recovered, which naturally motivates the use of machine learning techniques. We propose an image restoration algorithm based on the deep neural network framework where the denoising auto-encoders are stacked building multiple layers. The denoising auto-encoder is a variant of a classical auto-encoder that takes an input data and maps it to a hidden representation through a deterministic mapping using a non-linear activation function. The latent representation is then mapped back into a reconstruction the size of which is the same as the size of the input data. The reconstruction error can be measured by the traditional squared error assuming the residual follows a normal distribution. In addition to the designed loss function, an effective regularization scheme using residual-driven dropout determined based on the gradient at each layer. The optimal weights are computed by the classical stochastic gradient descent algorithm combined with the back-propagation algorithm. In our algorithm, we initially decompose an input image into its intrinsic representation and the nuisance factors including artefacts based on the classical Total Variation problem that can be efficiently optimized by the convex optimization algorithm such as primal-dual method. The intrinsic forms of the input images are provided to the deep denosing auto-encoders with their original forms in the training phase. In the testing phase, a given image is first decomposed into the intrinsic form and then provided to the trained network to obtain its reconstruction. We apply our algorithm to the restoration of the corrupted CT images by the artefacts. It is shown that our algorithm improves the readability and enhances the anatomical and pathological properties of the object. The quantitative evaluation is performed in terms of the PSNR, and the qualitative evaluation provides significant improvement in reading images despite degrading artefacts. The experimental results indicate the potential of our algorithm as a prior solution to the image interpretation tasks in a variety of medical imaging applications. This work was supported by the MISP(Ministry of Science and ICT), Korea, under the National Program for Excellence in SW (20170001000011001) supervised by the IITP(Institute for Information and Communications Technology Promotion).

Keywords: auto-encoder neural network, CT image artefact, deep learning, intrinsic image representation, noise reduction, total variation

Procedia PDF Downloads 173

Authors: Zhidong Zhang, Yingchen Yang

Abstract:

In this study, a diagnostic assessment model for Mastery Engineering Learning was established based on a group of undergraduate students who studied in an engineering course. A diagnostic assessment model can examine both students' learning process and report achievement results. One very unique characteristic is that the diagnostic assessment model can recognize the errors and anything blocking students in their learning processes. The feedback is provided to help students to know how to solve the learning problems with alternative strategies and help the instructor to find alternative pedagogical strategies in the instructional designs. Dynamics is a core course in which is a common course being shared by several engineering programs. This course is a very challenging for engineering students to solve the problems. Thus knowledge acquisition and problem-solving skills are crucial for student success. Therefore, developing an effective and valid assessment model for student learning are of great importance. Diagnostic assessment is such a model which can provide effective feedback for both students and instructor in the mastery of engineering learning.

Keywords: diagnostic assessment, mastery learning, engineering, bayesian network model, learning processes

Procedia PDF Downloads 139

Authors: Revoti Prasad Bora, Saurabh Yadav, Nikita Katyal

Abstract:

In the context of educational data mining, the use case of extracting information from images containing both text and diagrams is of high importance. Hence, document analysis requires the extraction of diagrams from such images and processes the text and diagrams separately. To the author’s best knowledge, none among plenty of approaches for extracting tables, figures, etc., suffice the need for real-time processing with high accuracy as needed in multiple applications. In the education domain, diagrams can be of varied characteristics viz. line-based i.e. geometric diagrams, chemical bonds, mathematical formulas, etc. There are two broad categories of approaches that try to solve similar problems viz. traditional computer vision based approaches and deep learning approaches. The traditional computer vision based approaches mainly leverage connected components and distance transform based processing and hence perform well in very limited scenarios. The existing deep learning approaches either leverage YOLO or faster-RCNN architectures. These approaches suffer from a performance-accuracy tradeoff. This paper proposes a U-Net based architecture that formulates the diagram extraction as a segmentation problem. The proposed method provides similar accuracy with a much faster extraction time as compared to the mentioned state-of-the-art approaches. Further, the segmentation mask in this approach allows the extraction of diagrams of irregular shapes.

Keywords: computer vision, deep-learning, educational data mining, faster-RCNN, figure extraction, image segmentation, real-time document analysis, text extraction, U-Net, YOLO

Procedia PDF Downloads 110

Authors: Nguyen J. M., Lucas G., Ruan S., Digonnet H., Antonioli D.

Abstract:

Introduction: Deep Learning (DL) is a very powerful tool for analyzing image data. But for tabular data, it cannot compete with machine learning methods like XGBoost. The research question becomes: can tabular data be transformed into images that can be analyzed by simple CNNs (Convolutional Neuron Networks)? Will DL be the absolute tool for data classification? All current solutions consist in repositioning the variables in a 2x2 matrix using their correlation proximity. In doing so, it obtains an image whose pixels are the variables. We implement a technology, DeepNIC, that offers the possibility of obtaining an image for each variable, which can be analyzed by simple CNNs. Material and method: The 'ROP' (Regression OPtimized) model is a binary and atypical decision tree whose nodes are managed by a new artificial neuron, the Neurop. By positioning an artificial neuron in each node of the decision trees, it is possible to make an adjustment on a theoretically infinite number of variables at each node. From this new decision tree whose nodes are artificial neurons, we created the concept of a 'Random Forest of Perfect Trees' (RFPT), which disobeys Breiman's concepts by assembling very large numbers of small trees with no classification errors. From the results of the RFPT, we developed a family of 10 statistical information criteria, Nguyen Information Criterion (NICs), which evaluates in 3 dimensions the predictive quality of a variable: Performance, Complexity and Multiplicity of solution. A NIC is a probability that can be transformed into a grey level. The value of a NIC depends essentially on 2 super parameters used in Neurops. By varying these 2 super parameters, we obtain a 2x2 matrix of probabilities for each NIC. We can combine these 10 NICs with the functions AND, OR, and XOR. The total number of combinations is greater than 100,000. In total, we obtain for each variable an image of at least 1166x1167 pixels. The intensity of the pixels is proportional to the probability of the associated NIC. The color depends on the associated NIC. This image actually contains considerable information about the ability of the variable to make the prediction of Y, depending on the presence or absence of other variables. A basic CNNs model was trained for supervised classification. Results: The first results are impressive. Using the GSE22513 public data (Omic data set of markers of Taxane Sensitivity in Breast Cancer), DEEPNic outperformed other statistical methods, including XGBoost. We still need to generalize the comparison on several databases. Conclusion: The ability to transform any tabular variable into an image offers the possibility of merging image and tabular information in the same format. This opens up great perspectives in the analysis of metadata.

Keywords: tabular data, CNNs, NICs, DeepNICs, random forest of perfect trees, classification

Procedia PDF Downloads 90

Authors: Sumit Sen, Sohan Khedekar, Umang Shinde, Shivam Bhargava

Abstract:

This study has been undertaken to investigate whether the deep learning models are able to predict the future stock prices by training the model with the historical stock price data. Since this work required time series analysis, various models are present today to perform time series analysis such as Recurrent Neural Network LSTM, ARIMA and Facebook Prophet. Applying these models the movement of stock price of stocks are predicted and also tried to provide the future prediction of the stock price of a stock. Final product will be a stock price prediction web application that is developed for providing the user the ease of analysis of the stocks and will also provide the predicted stock price for the next seven days.

Keywords: Autoregressive Integrated Moving Average, Deep Learning, Long Short Term Memory, Time-series

Procedia PDF Downloads 120

Authors: Mehwish Asghar

Abstract:

Breast Cancer (BC) is a common type of cancer among women. Its screening is usually performed using different imaging modalities such as magnetic resonance imaging, mammogram, X-ray, CT, etc. Among these modalities’ mammogram is considered a powerful tool for diagnosis and screening of breast cancer. Sophisticated machine learning approaches have shown promising results in complementing human diagnosis. Generally, machine learning methods can be divided into two major classes: one is Radiomics analysis (RA), where image features are extracted manually; and the other one is the concept of convolutional neural networks (CNN), in which the computer learns to recognize image features on its own. This research aims to improve the incidence of early detection, thus reducing the mortality rate caused by breast cancer through the latest advancements in computer science, in general, and machine learning, in particular. It has also been aimed to ease the burden of doctors by improving and automating the process of breast cancer detection. This research is related to a relative analysis of different techniques for the implementation of different models for detecting and classifying breast cancer. The main goal of this research is to provide a detailed view of results and performances between different techniques. The purpose of this paper is to explore the potential of a convolutional neural network (CNN) w.r.t feature extractor and as a classifier. Also, in this research, it has been aimed to add the module of Radiomics for comparison of its results with deep learning techniques.

Keywords: breast cancer (BC), machine learning (ML), convolutional neural network (CNN), radionics, magnetic resonance imaging, artificial intelligence

Procedia PDF Downloads 198

Authors: Gorkem Algan, Ilkay Ulusoy, Saban Gonul, Banu Turgut, Berker Bakbak

Abstract:

Computer vision systems recently made a big leap thanks to deep neural networks. However, these systems require correctly labeled large datasets in order to be trained properly, which is very difficult to obtain for medical applications. Two main reasons for label noise in medical applications are the high complexity of the data and conflicting opinions of experts. Moreover, medical imaging datasets are commonly tiny, which makes each data very important in learning. As a result, if not handled properly, label noise significantly degrades the performance. Therefore, a label-noise-robust learning algorithm that makes use of the meta-learning paradigm is proposed in this article. The proposed solution is tested on retinopathy of prematurity (ROP) dataset with a very high label noise of 68%. Results show that the proposed algorithm significantly improves the classification algorithm's performance in the presence of noisy labels.

Keywords: deep learning, label noise, robust learning, meta-learning, retinopathy of prematurity

Procedia PDF Downloads 135

Authors: Leila Keshavarz Afshar, Hedieh Sajedi

Abstract:

Estimation of biological brain age from MR images is a topic that has been much addressed in recent years due to the importance it attaches to early diagnosis of diseases such as Alzheimer's. In this paper, we use a 3D Convolutional Neural Network (CNN) to provide a method for estimating the biological age of the brain. The 3D-CNN model is trained by MRI data that has been normalized. In addition, to reduce computation while saving overall performance, some effectual slices are selected for age estimation. By this method, the biological age of individuals using selected normalized data was estimated with Mean Absolute Error (MAE) of 4.82 years.

Keywords: brain age estimation, biological age, 3D-CNN, deep learning, T1-weighted image, SPM, preprocessing, MRI, canny, gray matter

Procedia PDF Downloads 127

Authors: Hussain Sajid, Jung-Hun Shin, Kum-Won Cho

Abstract:

Image super-resolution is the most common computer vision problem with many important applications. Generative adversarial networks (GANs) have promoted remarkable advances in single-image super-resolution (SR) by recovering photo-realistic images. However, high memory requirements of GAN-based SR (mainly generators) lead to performance degradation and increased energy consumption, making it difficult to implement it onto resource-constricted devices. To relieve such a problem, In this paper, we introduce an optimized and highly efficient architecture for SR-GAN (generator) model by utilizing model compression techniques such as Knowledge Distillation and pruning, which work together to reduce the storage requirement of the model also increase in their performance. Our method begins with distilling the knowledge from a large pre-trained model to a lightweight model using different loss functions. Then, iterative weight pruning is applied to the distilled model to remove less significant weights based on their magnitude, resulting in a sparser network. Knowledge Distillation reduces the model size by 40%; pruning then reduces it further by 18%. To accelerate the learning process, we employ the Horovod framework for distributed training on a cluster of 2 nodes, each with 8 GPUs, resulting in improved training performance and faster convergence. Experimental results on various benchmarks demonstrate that the proposed compressed model significantly outperforms state-of-the-art methods in terms of peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and image quality for x4 super-resolution tasks.

Keywords: single-image super-resolution, generative adversarial networks, knowledge distillation, pruning

Procedia PDF Downloads 61

Authors: Zhou Hong, Gu Xiao-Qing, Lıu Hong-Jiao, Leng Jing

Abstract:

Seamless learning is becoming a research hotspot in recent years, and the emerging of micro-lectures, flipped classroom has strengthened the development of seamless learning. Based on the characteristics of the seamless learning across time and space and the course structure of the flipped classroom, and the theories of language learning, we put forward the language learning model which can support ‘seamless flipped’ environment (abbreviated as ‘S-F’). Meanwhile, the characteristics of the ‘S-F’ learning environment, the corresponding framework construction and the activity design of diversified corpora were introduced. Moreover, a language learning app named ‘Daily Speaking’ was developed to facilitate the practice of the language learning model in ‘S-F’ environment. In virtue of the learning case of Shanghai language, the rationality and feasibility of this framework were examined, expecting to provide a reference for the design of ‘S-F’ learning in different situations.

Keywords: seamless learning, flipped classroom, seamless-flipped environment, language learning model

Procedia PDF Downloads 165

Authors: Jiahe Liu, HongyangYu, Feng Qian, Miao Luo

Abstract:

This study aimed to address the problem of improving the confidence of key points by fusing multi-view information, thereby estimating human posture more accurately. We first obtained multi-view image information and then used the MvP algorithm to fuse this multi-view information together to obtain a set of high-confidence human key points. We used these as the input for the Spatio-Temporal Graph Convolution (ST-GCN). ST-GCN is a deep learning model used for processing spatio-temporal data, which can effectively capture spatio-temporal relationships in video sequences. By using the MvP algorithm to fuse multi-view information and inputting it into the spatio-temporal graph convolution model, this study provides an effective method to improve the accuracy of human posture estimation and provides strong support for further research and application in related fields.

Keywords: multi-view, pose estimation, ST-GCN, joint fusion

Procedia PDF Downloads 46

Authors: Xiuying Wang, Changliang Li, Bo Xu

Abstract:

A fixed-length feature vector is required for many machine learning algorithms in NLP field. Word embeddings have been very successful at learning lexical information. However, they cannot capture the compositional meaning of sentences, which prevents them from a deeper understanding of language. In this paper, we introduce a novel hierarchical tree long short-term memory (HTLSTM) model that learns vector representations for sentences of arbitrary syntactic type and length. We propose to split one sentence into three hierarchies: short phrase, long phrase and full sentence level. The HTLSTM model gives our algorithm the potential to fully consider the hierarchical information and long-term dependencies of language. We design the experiments on both English and Chinese corpus to evaluate our model on sentiment analysis task. And the results show that our model outperforms several existing state of the art approaches significantly.

Keywords: deep learning, hierarchical tree long short-term memory, sentence representation, sentiment analysis

Procedia PDF Downloads 335

Authors: Muneeb Ullah, Daishihan, Xiadong Young

Abstract:

Deep learning plays a crucial role in identifying COVID-19 and preventing its spread. To improve the accuracy of COVID-19 diagnoses, it is important to have access to a sufficient number of training images of CXRs (chest X-rays) depicting the disease. However, there is currently a shortage of such images. To address this issue, this paper introduces COVID-19 GAN, a model that uses generative adversarial networks (GANs) to generate realistic CXR images of COVID-19, which can be used to train identification models. Initially, a generator model is created that uses digressive channels to generate images of CXR scans for COVID-19. To differentiate between real and fake disease images, an efficient discriminator is developed by combining the dense connectivity strategy and instance normalization. This approach makes use of their feature extraction capabilities on CXR hazy areas. Lastly, the deep regret gradient penalty technique is utilized to ensure stable training of the model. With the use of 4,062 grape leaf disease images, the Leaf GAN model successfully produces 8,124 COVID-19 CXR images. The COVID-19 GAN model produces COVID-19 CXR images that outperform DCGAN and WGAN in terms of the Fréchet inception distance. Experimental findings suggest that the COVID-19 GAN-generated CXR images possess noticeable haziness, offering a promising approach to address the limited training data available for COVID-19 model training. When the dataset was expanded, CNN-based classification models outperformed other models, yielding higher accuracy rates than those of the initial dataset and other augmentation techniques. Among these models, ImagNet exhibited the best recognition accuracy of 99.70% on the testing set. These findings suggest that the proposed augmentation method is a solution to address overfitting issues in disease identification and can enhance identification accuracy effectively.

Keywords: classification, deep learning, medical images, CXR, GAN.

Procedia PDF Downloads 60

Authors: Roshani S. Golhar, Neerajkumar S. Sathawane, Snehal Dongre

Abstract:

Background and objectives: As the, cardiovascular illnesses increasing and becoming cause of mortality worldwide, killing around lot of people each year. Arrhythmia is a type of cardiac illness characterized by a change in the linearity of the heartbeat. The goal of this study is to develop novel deep learning algorithms for successfully interpreting arrhythmia using a single second segment. Because the ECG signal indicates unique electrical heart activity across time, considerable changes between time intervals are detected. Such variances, as well as the limited number of learning data available for each arrhythmia, make standard learning methods difficult, and so impede its exaggeration. Conclusions: The proposed method was able to outperform several state-of-the-art methods. Also proposed technique is an effective and convenient approach to deep learning for heartbeat interpretation, that could be probably used in real-time healthcare monitoring systems

Keywords: electrocardiogram, ECG classification, neural networks, convolutional neural networks, portable document format

Procedia PDF Downloads 54

Authors: Gong Zhilin, Jing Yang, Jian Yin

Abstract:

The purpose of this paper is to develop a novel system for credit card fraud detection based on sequential modeling of data using hybrid deep learning models. The projected model encapsulates five major phases are pre-processing, imbalance-data handling, feature extraction, optimal feature selection, and fraud detection with an ensemble classifier. The collected raw data (input) is pre-processed to enhance the quality of the data through alleviation of the missing data, noisy data as well as null values. The pre-processed data are class imbalanced in nature, and therefore they are handled effectively with the K-means clustering-based SMOTE model. From the balanced class data, the most relevant features like improved Principal Component Analysis (PCA), statistical features (mean, median, standard deviation) and higher-order statistical features (skewness and kurtosis). Among the extracted features, the most optimal features are selected with the Self-improved Arithmetic Optimization Algorithm (SI-AOA). This SI-AOA model is the conceptual improvement of the standard Arithmetic Optimization Algorithm. The deep learning models like Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and optimized Quantum Deep Neural Network (QDNN). The LSTM and CNN are trained with the extracted optimal features. The outcomes from LSTM and CNN will enter as input to optimized QDNN that provides the final detection outcome. Since the QDNN is the ultimate detector, its weight function is fine-tuned with the Self-improved Arithmetic Optimization Algorithm (SI-AOA).

Keywords: credit card, data mining, fraud detection, money transactions

Procedia PDF Downloads 111

Authors: Surajit Chakrabarty, Piyush Chauhan, Subhasis Panda, Sujoy Bhattacharya

Abstract:

A large proportion of roads in India are not well maintained as per the laid down public safety guidelines leading to loss of direction control and fatal accidents. We propose a technique to detect speed breakers and potholes using mobile sensor data captured from multiple vehicles and provide a profile of the road. This would, in turn, help in monitoring roads and revolutionize digital maps. Incorporating randomness in the model formulation for detection of speed breakers and potholes is crucial due to substantial heterogeneity observed in data obtained using a mobile application from multiple vehicles driven by different drivers. This is accomplished with Hidden Markov Models, whose hidden state sequence is found for each time step given the observables sequence, and are then fed as input to LSTM network with peephole connections. A precision score of 0.96 and 0.63 is obtained for classifying bumps and potholes, respectively, a significant improvement from the machine learning based models. Further visualization of bumps/potholes is done by converting time series to images using Markov Transition Fields where a significant demarcation among bump/potholes is observed.

Keywords: deep learning, hidden Markov model, pothole, speed breaker

Procedia PDF Downloads 122

Authors: Mohammad Shameel, Peter K. K. Loh, James H. Ng

Abstract:

Cybercrime is a new category of crime which poses a different challenge for crime investigators and incident responders. Attackers can mask their identities using a suite of tools and with the help of the deep web, which makes them difficult to track down. Scouring the deep web manually takes time and is inefficient. There is a growing need for a tool to scour the deep web to obtain useful evidence or intel automatically. In this paper, we will explain the background and motivation behind the research, present a survey of existing research on related tools, describe the design of our own crime/threat intelligence hunting tool prototype, demonstrate its capability with some test cases and lastly, conclude with proposals for future enhancements.

Keywords: cybercrime, deep web, threat intelligence, web crawler

Procedia PDF Downloads 148

Authors: Gulfam Haider, sana danish

Abstract:

Convolutional Neural Networks (CNNs) have emerged as powerful tools for image classification, and the choice of optimizers profoundly affects their performance. The study of optimizers and their adaptations remains a topic of significant importance in machine learning research. While numerous studies have explored and advocated for various optimizers, the efficacy of these optimization techniques is still subject to scrutiny. This work aims to address the challenges surrounding the effectiveness of optimizers by conducting a comprehensive analysis and evaluation. The primary focus of this investigation lies in examining the performance of different optimizers when employed in conjunction with the popular activation function, Rectified Linear Unit (ReLU). By incorporating ReLU, known for its favorable properties in prior research, the aim is to bolster the effectiveness of the optimizers under scrutiny. Specifically, we evaluate the adjustment of these optimizers with both the original Softmax activation function and the modified ReLU activation function, carefully assessing their impact on overall performance. To achieve this, a series of experiments are conducted using a well-established benchmark dataset for image classification tasks, namely the Canadian Institute for Advanced Research dataset (CIFAR-10). The selected optimizers for investigation encompass a range of prominent algorithms, including Adam, Root Mean Squared Propagation (RMSprop), Adaptive Learning Rate Method (Adadelta), Adaptive Gradient Algorithm (Adagrad), and Stochastic Gradient Descent (SGD). The performance analysis encompasses a comprehensive evaluation of the classification accuracy, convergence speed, and robustness of the CNN models trained with each optimizer. Through rigorous experimentation and meticulous assessment, we discern the strengths and weaknesses of the different optimization techniques, providing valuable insights into their suitability for image classification tasks. By conducting this in-depth study, we contribute to the existing body of knowledge surrounding optimizers in CNNs, shedding light on their performance characteristics for image classification. The findings gleaned from this research serve to guide researchers and practitioners in making informed decisions when selecting optimizers and activation functions, thus advancing the state-of-the-art in the field of image classification with convolutional neural networks.

Keywords: deep neural network, optimizers, RMsprop, ReLU, stochastic gradient descent

Procedia PDF Downloads 96

Authors: Snehal G. Teli, R. J. Shelke

Abstract:

CNN and MultiUNet models are the framework for the proposed method for enhancing and reconstructing underwater images. Multiscale merging of features and regeneration are both performed by the MultiUNet. CNN collects relevant features. Extensive tests on benchmark datasets show that the proposed strategy performs better than the latest methods. As a result of this work, underwater images can be represented and interpreted in a number of underwater applications with greater clarity. This strategy will advance underwater exploration and marine research by enhancing real-time underwater image processing systems, underwater robotic vision, and underwater surveillance.

Keywords: convolutional neural network, image enhancement, machine learning, multiunet, underwater images

Procedia PDF Downloads 50

Authors: Hui Wei, Zheng Dong

Abstract:

Object recognition has long been a challenging task in computer vision. Yet the human brain, with the ability to rapidly and accurately recognize visual stimuli, manages this task effortlessly. In the past decades, advances in neuroscience have revealed some neural mechanisms underlying visual processing. In this paper, we present a novel model inspired by the visual pathway in primate brains. This multi-layer neural network model imitates the hierarchical convergent processing mechanism in the visual pathway. We show that local image features generated by this model exhibit robust discrimination and even better generalization ability compared with some existing image descriptors. We also demonstrate the application of this model in an object recognition task on image data sets. The result provides strong support for the potential of this model.

Keywords: biological model, feature extraction, multi-layer neural network, object recognition

Procedia PDF Downloads 521

Authors: Ting Lu, Ya-Ru Diao, Fei Fan, Ye Xue, Lei Shi, Xian-e Tang, Meng-jun Zhan, Zhen-hua Deng

Abstract:

Accurate adult age estimation (AAE) is a significant and challenging task in forensic and archeology fields. Attempts have been made to explore optimal adult age metrics, and the rib is considered a potential age marker. The traditional way is to extract age-related features designed by experts from macroscopic or radiological images followed by classification or regression analysis. Those results still have not met the high-level requirements for practice, and the limitation of using feature design and manual extraction methods is loss of information since the features are likely not designed explicitly for extracting information relevant to age. Deep learning (DL) has recently garnered much interest in imaging learning and computer vision. It enables learning features that are important without a prior bias or hypothesis and could be supportive of AAE. This study aimed to develop DL models for AAE based on CT images and compare their performance to the manual visual scoring method. Chest CT data were reconstructed using volume rendering (VR). Retrospective data of 2500 patients aged 20.00-69.99 years were obtained between December 2019 and September 2021. Five-fold cross-validation was performed, and datasets were randomly split into training and validation sets in a 4:1 ratio for each fold. Before feeding the inputs into networks, all images were augmented with random rotation and vertical flip, normalized, and resized to 224×224 pixels. ResNeXt was chosen as the DL baseline due to its advantages of higher efficiency and accuracy in image classification. Mean absolute error (MAE) was the primary parameter. Independent data from 100 patients acquired between March and April 2022 were used as a test set. The manual method completely followed the prior study, which reported the lowest MAEs (5.31 in males and 6.72 in females) among similar studies. CT data and VR images were used. The radiation density of the first costal cartilage was recorded using CT data on the workstation. The osseous and calcified projections of the 1 to 7 costal cartilages were scored based on VR images using an eight-stage staging technique. According to the results of the prior study, the optimal models were the decision tree regression model in males and the stepwise multiple linear regression equation in females. Predicted ages of the test set were calculated separately using different models by sex. A total of 2600 patients (training and validation sets, mean age=45.19 years±14.20 [SD]; test set, mean age=46.57±9.66) were evaluated in this study. Of ResNeXt model training, MAEs were obtained with 3.95 in males and 3.65 in females. Based on the test set, DL achieved MAEs of 4.05 in males and 4.54 in females, which were far better than the MAEs of 8.90 and 6.42 respectively, for the manual method. Those results showed that the DL of the ResNeXt model outperformed the manual method in AAE based on CT reconstruction of the costal cartilage and the developed system may be a supportive tool for AAE.

Keywords: forensic anthropology, age determination by the skeleton, costal cartilage, CT, deep learning

Procedia PDF Downloads 53

Authors: Konstantinos Perifanos, Eirini Florou, Dionysis Goutsos

Abstract:

This paper presents and benchmarks a number of end-to-end Deep Learning based models for metaphor detection in Greek. We combine Convolutional Neural Networks and Recurrent Neural Networks with representation learning to bear on the metaphor detection problem for the Greek language. The models presented achieve exceptional accuracy scores, significantly improving the previous state-of-the-art results, which had already achieved accuracy 0.82. Furthermore, no special preprocessing, feature engineering or linguistic knowledge is used in this work. The methods presented achieve accuracy of 0.92 and F-score 0.92 with Convolutional Neural Networks (CNNs) and bidirectional Long Short Term Memory networks (LSTMs). Comparable results of 0.91 accuracy and 0.91 F-score are also achieved with bidirectional Gated Recurrent Units (GRUs) and Convolutional Recurrent Neural Nets (CRNNs). The models are trained and evaluated only on the basis of training tuples, the related sentences and their labels. The outcome is a state-of-the-art collection of metaphor detection models, trained on limited labelled resources, which can be extended to other languages and similar tasks.

Keywords: metaphor detection, deep learning, representation learning, embeddings

Procedia PDF Downloads 130

Authors: Subodh Chandra Shakya, Badal Shrestha, Suni Thapa, Ashutosh Chauhan, Saugat Adhikari

Abstract:

To tackle our problem of clothes identification, we used different architectures of Convolutional Neural Networks. Among different architectures, the outcome from Inception ResNet V2 and MobileNet V2 seemed promising. On comparison of the metrices, we observed that the Inception ResNet V2 slightly outperforms MobileNet V2 for this purpose. So this paper of ours proposes the cloth identifier using Inception ResNet V2 and also contains the comparison between the outcome of ResNet V2 and MobileNet V2. The document here contains the results and findings of the research that we performed on the DeepFashion Dataset. To improve the dataset, we used different image preprocessing techniques like image shearing, image rotation, and denoising. The whole experiment was conducted with the intention of testing the efficiency of convolutional neural networks on cloth identification so that we could develop a reliable system that is good enough in identifying the clothes worn by the users. The whole system can be integrated with some kind of recommendation system.

Keywords: inception ResNet, convolutional neural net, deep learning, confusion matrix, data augmentation, data preprocessing

Procedia PDF Downloads 166

Authors: Munazza A. Mirza, Khawar Khurshid

Abstract:

Individuals are generally associated with different learning styles, which have been explored extensively in recent past. The learning styles refer to the potential of an individual by which s/he can easily comprehend and retain information. Among various learning style models, VARK is the most accepted model which categorizes the learners with respect to their sensory characteristics. Based on the number of preferred learning modes, the learners can be categorized as uni-modal, bi-modal, tri-modal, or quad/multi-modal. Although there is a prevalent belief in the learning styles, however, the model is not being frequently and effectively utilized in the higher education. This research describes the identification model to validate teacher’s didactic practice and student’s performance linkage with the learning styles. The identification model is recommended to check the effective application and evaluation of the various learning styles. The proposed model is a guideline to effectively implement learning styles inventory in order to ensure that it will validate performance linkage with learning styles. If performance is linked with learning styles, this may help eradicate the distrust on learning style theory. For this purpose, a comprehensive study was conducted to compare and understand how VARK inventory model is being used to identify learning preferences and their correlation with learner’s performance. A comparative analysis of the findings of these studies is presented to understand the learning styles of tertiary students in various disciplines. It is concluded with confidence that the learning styles of students cannot be associated with any specific discipline. Furthermore, there is not enough empirical proof to link performance with learning styles.

Keywords: learning style, VARK, sensory preferences, identification model, didactic practices

Procedia PDF Downloads 248

Authors: Richard O’Riordan, Saritha Unnikrishnan

Abstract:

Autonomous Vehicles (AVs) are incorporating an increasing number of ADAS features, including automated lane-keeping systems. In recent years, many research papers into lane detection algorithms have been published, varying from computer vision techniques to deep learning methods. The transition from lower levels of autonomy defined in the SAE framework and the progression to higher autonomy levels requires increasingly complex models and algorithms that must be highly reliable in their operation and functionality capacities. Furthermore, these algorithms have no room for error when operating at high levels of autonomy. Although the current research details existing computer vision and deep learning algorithms and their methodologies and individual results, the research also details challenges faced by the algorithms and the resources needed to operate, along with shortcomings experienced during their detection of lanes in certain weather and lighting conditions. This paper will explore these shortcomings and attempt to implement a lane detection algorithm that could be used to achieve improvements in AV lane detection systems. This paper uses a pre-trained LaneNet model to detect lane or non-lane pixels using binary segmentation as the base detection method using an existing dataset BDD100k followed by a custom dataset generated locally. The selected roads will be modern well-laid roads with up-to-date infrastructure and lane markings, while the second road network will be an older road with infrastructure and lane markings reflecting the road network's age. The performance of the proposed method will be evaluated on the custom dataset to compare its performance to the BDD100k dataset. In summary, this paper will use Transfer Learning to provide a fast and robust lane detection algorithm that can handle various road conditions and provide accurate lane detection.

Keywords: ADAS, autonomous vehicles, deep learning, LaneNet, lane detection

Procedia PDF Downloads 79

Authors: Chinlun Lai, Lunjyh Jiang

Abstract:

Due to the heavy burden and pressure of caring for infants, an integrated automatic baby watching system based on IoT smart sensing and deep learning machine vision techniques is proposed in this paper. By monitoring infant body conditions such as heartbeat, breathing, body temperature, sleeping posture, as well as the surrounding conditions such as dangerous/sharp objects, light, noise, humidity and temperature, the proposed system can analyze and predict the obvious/potential dangerous conditions according to observed data and then adopt suitable actions in real time to protect the infant from harm. Thus, reducing the burden of the caregiver and improving safety efficiency of the caring work. The experimental results show that the proposed system works successfully for the infant care work and thus can be implemented in various life fields practically.

Keywords: baby care system, Internet of Things, deep learning, machine vision

Procedia PDF Downloads 208

Authors: Vidhu Mitha Goutham, Ruma Mukherjee

Abstract:

The low availability of well-trained, unlimited, dynamic-access models for specific languages makes it hard for corporate users to adopt quick translation techniques and incorporate them into product solutions. As translation tasks increasingly require a dynamic sequence learning curve; stable, cost-free opensource models are scarce. We survey and compare current translation techniques and propose a modified sequence to sequence model repurposed with attention techniques. Sequence learning using an encoder-decoder model is now paving the path for higher precision levels in translation. Using a Convolutional Neural Network (CNN) encoder and a Recurrent Neural Network (RNN) decoder background, we use Fairseq tools to produce an end-to-end bilingually trained Spanish-English machine translation model including source language detection. We acquire competitive results using a duo-lingo-corpus trained model to provide for prospective, ready-made plug-in use for compound sentences and document translations. Our model serves a decent system for large, organizational data translation needs. While acknowledging its shortcomings and future scope, it also identifies itself as a well-optimized deep neural network model and solution.

Keywords: attention, encoder-decoder, Fairseq, Seq2Seq, Spanish, translation

Procedia PDF Downloads 155

Authors: Riaan Kleyn

Abstract:

Mass wine production companies across the globe are provided with grapes from winegrowers that predominantly utilize mechanical harvesting machines to harvest wine grapes. Mechanical harvesting accelerates the rate at which grapes are harvested, allowing grapes to be delivered faster to meet the demands of wine cellars. The disadvantage of the mechanical harvesting method is the inclusion of material-other-than-grapes (MOG) in the harvested wine grape loads arriving at the cellar which degrades the quality of wine that can be produced. Currently, wine cellars do not have a method to determine the amount of MOG present within wine grape loads. This paper seeks to find an optimal computer vision method capable of detecting the amount of MOG within a wine grape load. A MOG detection method will encourage winegrowers to deliver MOG-free wine grape loads to avoid penalties which will indirectly enhance the quality of the wine to be produced. Traditional image segmentation methods were compared to deep learning segmentation methods based on images of wine grape loads that were captured at a wine cellar. The Mask R-CNN model with a ResNet-50 convolutional neural network backbone emerged as the optimal method for this study to determine the amount of MOG in an image of a wine grape load. Furthermore, a statistical analysis was conducted to determine how the MOG on the surface of a grape load relates to the mass of MOG within the corresponding grape load.

Keywords: computer vision, wine grapes, machine learning, machine harvested grapes

Procedia PDF Downloads 68