Search results for: Scale-Invariant Feature Transformation (SIFT)

Authors: Pengfei Meng, Shuangcheng Jia, Qian Li

Abstract:

We proposed a real-time high-precision semantic segmentation network based on a multi-resolution feature fusion module, the auxiliary feature extracting module, upsampling module, and atrous spatial pyramid pooling (ASPP) module. We designed a feature fusion structure, which is integrated with sufficient features of different resolutions. We also studied the effect of side-branch structure on the network and made discoveries. Based on the discoveries about the side-branch of the network structure, we used a side-branch auxiliary feature extraction layer in the network to improve the effectiveness of the network. We also designed upsampling module, which has better results than the original upsampling module. In addition, we also re-considered the locations and number of atrous spatial pyramid pooling (ASPP) modules and modified the network structure according to the experimental results to further improve the effectiveness of the network. The network presented in this paper takes the backbone network of Bisenetv2 as a basic network, based on which we constructed a network structure on which we made improvements. We named this network deep multiple-resolution bilateral networks for real-time, referred to as DMBR-Net. After experimental testing, our proposed DMBR-Net network achieved 81.2% mIoU at 119FPS on the Cityscapes validation dataset, 80.7% mIoU at 109FPS on the CamVid test dataset, 29.9% mIoU at 78FPS on the COCOStuff test dataset. Compared with all lightweight real-time semantic segmentation networks, our network achieves the highest accuracy at an appropriate speed.

Keywords: multi-resolution feature fusion, atrous convolutional, bilateral networks, pyramid pooling

Procedia PDF Downloads 125

Authors: Manojkumar Sabanayagam

Abstract:

The Riemann hypothesis is an unsolved millennium problem and the search for a solution to the Riemann hypothesis is to study the pattern of prime number distribution. The scope of this paper is to identify the solution for the critical strip and the critical line axis, which has the non-trivial zero solutions using complex plane functions. The Riemann graphical plot is constructed using a linear complex variable function (X+iY) and is applicable only when X>1. But the investigation shows that complex variable behavior has two zones. The first zone is the transformation zone, where the definition of the complex plane should be a non-linear variable which is the critical strip zone in the graph (X=0 to 1). The second zone is the transformed zone (X>1) defined using linear variables conventionally. This paper deals with the Non-linear function in the transformation zone derived using cosine and sinusoidal time lag w.r.t imaginary number ‘i’. The alternate complex variable (Cosθ+i Sinθ) is used to understand the variables in the critical strip zone. It is concluded that the non-trivial zeros present in the Real part 0.5 are because the linear function is not the correct approach in the critical strip. This paper provides the solution to Reimann's hypothesis.

Keywords: Reimann hypothesis, critical strip, complex plane, transformation zone

Procedia PDF Downloads 193

Authors: Zainab Mutlaq Ibrahim, Mohamed Bader-El-Den, Mihaela Cocea

Abstract:

Students' textual feedback can hold unique patterns and useful information about learning process, it can hold information about advantages and disadvantages of teaching methods, assessment components, facilities, and other aspects of teaching. The results of analysing such a feedback can form a key point for institutions’ decision makers to advance and update their systems accordingly. This paper proposes a data mining framework for analysing end of unit general textual feedback using part of speech feature (PoS) with four machine learning algorithms: support vector machines, decision tree, random forest, and naive bays. The proposed framework has two tasks: first, to use the above algorithms to build an optimal model that automatically classifies the whole data set into two subsets, one subset is tailored to assessment practices (assessment related), and the other one is the non-assessment related data. Second task to use the same algorithms to build an optimal model for whole data set, and the new data subsets to automatically detect their sentiment. The significance of this paper is to compare the performance of the above four algorithms using part of speech feature to the performance of the same algorithms using n-grams feature. The paper follows Knowledge Discovery and Data Mining (KDDM) framework to construct the classification and sentiment analysis models, which is understanding the assessment domain, cleaning and pre-processing the data set, selecting and running the data mining algorithm, interpreting mined patterns, and consolidating the discovered knowledge. The results of this paper experiments show that both models which used both features performed very well regarding first task. But regarding the second task, models that used part of speech feature has underperformed in comparison with models that used unigrams and bigrams.

Keywords: assessment, part of speech, sentiment analysis, student feedback

Procedia PDF Downloads 122

Authors: Jianghao Hu, Hongyu Wang

Abstract:

The effective and efficient human pose estimation algorithm is an important task for real-time human pose estimation on mobile devices. This paper proposes a light-weight human key points detection algorithm, Light-Weight Network for Real-Time Pose Estimation (LWPE). LWPE uses light-weight backbone network and depthwise separable convolutions to reduce parameters and lower latency. LWPE uses the feature pyramid network (FPN) to fuse the high-resolution, semantically weak features with the low-resolution, semantically strong features. In the meantime, with multi-scale prediction, the predicted result by the low-resolution feature map is stacked to the adjacent higher-resolution feature map to intermediately monitor the network and continuously refine the results. At the last step, the key point coordinates predicted in the highest-resolution are used as the final output of the network. For the key-points that are difficult to predict, LWPE adopts the online hard key points mining strategy to focus on the key points that hard predicting. The proposed algorithm achieves excellent performance in the single-person dataset selected in the AI (artificial intelligence) challenge dataset. The algorithm maintains high-precision performance even though the model only contains 3.9M parameters, and it can run at 225 frames per second (FPS) on the generic graphics processing unit (GPU).

Keywords: depthwise separable convolutions, feature pyramid network, human pose estimation, light-weight backbone

Procedia PDF Downloads 140

Authors: Saroj Bijarnia, Preety Singh

Abstract:

We propose a simple and effective biometrics system based on face verification across aging using a new variant of texture feature, Pyramid Binary Pattern. This employs Local Binary Pattern along with its hierarchical information. Dimension reduction of generated texture feature vector is done using Principal Component Analysis. Support Vector Machine is used for classification. Our proposed method achieves an accuracy of 92:24% and can be used in an automated age-invariant face verification system.

Keywords: biometrics, age invariant, verification, support vector machine

Procedia PDF Downloads 331

Authors: Aymen Bouzid, Jihen Ben Smida

Abstract:

In this paper, we propose to prove the importance of the application of blind source separation methods on remote sensing data in order to index multispectral images. The proposed method starts with Gabor Filtering and the application of a Blind Source Separation to get a more effective representation of the information contained on the observation images. After that, a feature vector is extracted from each image in order to index them. Experimental results show the superior performance of this approach.

Keywords: blind source separation, content based image retrieval, feature extraction multispectral, satellite images

Procedia PDF Downloads 386

Authors: Andre Peres Ramos, Franklin Cesar Flores

Abstract:

Colorization is the process of add colors to a monochromatic image or video. Usually, the process involves to segment the image in regions of interest and then apply colors to each one, for videos, this process is repeated for each frame, which makes it a tedious and time-consuming job. We propose a new assisted method for video colorization; the user only has to colorize one frame, and then the colors are propagated to following frames. The user can intervene at any time to correct eventual errors in color assignment. The method consists of to extract intensity and texture descriptors from the frames and then perform a feature matching to determine the best color for each segment. To reduce computation time and give a better spatial coherence we narrow the area of search and give weights for each feature to emphasize texture descriptors. To give a more natural result, we use an optimization algorithm to make the color propagation. Experimental results in several image sequences, compared to others existing methods, demonstrates that the proposed method perform a better colorization with less time and user interference.

Keywords: colorization, feature matching, texture descriptors, video segmentation

Procedia PDF Downloads 146

Authors: Lae-Jeong Park

Abstract:

The paper presents a method that utilizes figure-ground color segmentation to extract effective global feature in terms of false positive reduction in the head-shoulder detection. Conventional detectors that rely on local features such as HOG due to real-time operation suffer from false positives. Color cue in an input image provides salient information on a global characteristic which is necessary to alleviate the false positives of the local feature based detectors. An effective approach that uses figure-ground color segmentation has been presented in an effort to reduce the false positives in object detection. In this paper, an extended version of the approach is presented that adopts separate multipart foregrounds instead of a single prior foreground and performs the figure-ground color segmentation with each of the foregrounds. The multipart foregrounds include the parts of the head-shoulder shape and additional auxiliary foregrounds being optimized by a search algorithm. A classifier is constructed with the feature that consists of a set of the multiple resulting segmentations. Experimental results show that the presented method can discriminate more false positive than the single prior shape-based classifier as well as detectors with the local features. The improvement is possible because the presented approach can reduce the false positives that have the same colors in the head and shoulder foregrounds.

Keywords: pedestrian detection, color segmentation, false positive, feature extraction

Procedia PDF Downloads 260

Authors: Wesley Preßler, Lucie Schmidt

Abstract:

Digital transformation is playing an increasingly important role in the development of smart residential quarters. In order to accompany and steer this process and ultimately make the success of the transformation efforts measurable, it is helpful to use an appropriate maturity model. However, conventional maturity models for digital transformation focus primarily on the evaluation of processes and neglect the information and power imbalances between the stakeholders, which affects the validity of the results. The Multi-Generation Smart Community (mGeSCo) research project is developing an interdisciplinary maturity model that integrates the dimensions of digital literacy, interpretive patterns, and technology acceptance to address this gap. As part of the mGeSCo project, the technological development of selected dimensions in the Smart Quarter Jena-Lobeda (Germany) is being investigated. A specific maturity model, based on Cohen's Smart Cities Wheel, evaluates the central dimensions Working, Living, Housing and Caring. To improve the reliability and relevance of the maturity assessment, the factors Digital Literacy, Interpretive Patterns and Technology Acceptance are integrated into the developed model. The digital literacy dimension examines stakeholders' skills in using digital technologies, which influence their perception and assessment of technological maturity. Digital literacy is measured by means of surveys, interviews, and participant observation, using the European Commission's Digital Literacy Framework (DigComp) as a basis. Interpretations of digital technologies provide information about how individuals perceive technologies and ascribe meaning to them. However, these are not mere assessments, prejudices, or stereotyped perceptions but collective patterns, rules, attributions of meaning and the cultural repertoire that leads to these opinions and attitudes. Understanding these interpretations helps in assessing the overarching readiness of stakeholders to digitally transform a/their neighborhood. This involves examining people's attitudes, beliefs, and values about technology adoption, as well as their perceptions of the benefits and risks associated with digital tools. These insights provide important data for a holistic view and inform the steps needed to prepare individuals in the neighborhood for a digital transformation. Technology acceptance is another crucial factor for successful digital transformation to examine the willingness of individuals to adopt and use new technologies. Surveys or questionnaires based on Davis' Technology Acceptance Model can be used to complement interpretive patterns to measure neighborhood acceptance of digital technologies. Integrating the dimensions of digital literacy, interpretive patterns and technology acceptance enables the development of a roadmap with clear prerequisites for initiating a digital transformation process in the neighborhood. During the process, maturity is measured at different points in time and compared with changes in the aforementioned dimensions to ensure sustainable transformation. Participation, co-creation, and co-production are essential concepts for a successful and inclusive digital transformation in the neighborhood context. This interdisciplinary maturity model helps to improve the assessment and monitoring of sustainable digital transformation processes in smart residential quarters. It enables a more comprehensive recording of the factors that influence the success of such processes and supports the development of targeted measures to promote digital transformation in the neighborhood context.

Keywords: digital transformation, interdisciplinary, maturity model, neighborhood

Procedia PDF Downloads 60

Authors: Kamal Amarouche, Houda Benbrahim, Ismail Kassou

Abstract:

Nowadays, e-commerce shopping websites have experienced noticeable growth. These websites have gained consumers’ trust. After purchasing a product, many consumers share comments where opinions are usually embedded about the given product. Research on the automatic management of opinions that gives suggestions to potential consumers and portrays an image of the product to manufactures has been growing recently. After launching the product in the market, the reviews generated around it do not usually contain helpful information or generic opinions about this product (e.g. telephone: great phone...); in the sense that the product is still in the launching phase in the market. Within time, the product becomes old. Therefore, consumers perceive the advantages/ disadvantages about each specific product feature. Therefore, they will generate comments that contain their sentiments about these features. In this paper, we present an unsupervised method to extract different product features hidden in the opinions which influence its purchase, and that combines Time Weighting (TW) which depends on the time opinions were expressed with Term Frequency-Inverse Document Frequency (TF-IDF). We conduct several experiments using two different datasets about cell phones and hotels. The results show the effectiveness of our automatic feature extraction, as well as its domain independent characteristic.

Keywords: opinion mining, product feature extraction, sentiment analysis, SentiWordNet

Procedia PDF Downloads 381

Authors: Fatih Sogut

Abstract:

Developments in information technologies have caused significant changes in terms of radio and television broadcasting. With these changes in terms of production format, transmission techniques and service delivery, the distinction between traditional media and New Media has emerged. The viewer/listener, who was in a passive position before, is now in an active position and has a say in many matters, including content production. Visual and auditory data transfer has diversified and become easier thanks to the convergence phenomenon. These transformations and developments also affected one of the oldest electronic communication tools, radio. In this study, in order to adapt to the new era that emerged with the digital age, the change in radio broadcasting and the factors that led to this change were tried to be explained.

Keywords: Internet, radio broadcasting, digital transformation, Internet broadcasting

Procedia PDF Downloads 153

Authors: Nasar Aldian Ambark Shashoa

Abstract:

Classification with a dimension reduction based on Bayesian approach is proposed in this paper . The first step is to generate a sample (parameter) of fault-free mode class and faulty mode class. The second, in order to obtain good classification performance, a selection of important features is done with the discrete karhunen-loeve expansion. Next, the Bayes test for minimum error is used to classify the classes. Finally, the results for simulated data demonstrate the capabilities of the proposed procedure.

Keywords: analytical redundancy, fault detection, feature extraction, Bayesian approach

Procedia PDF Downloads 509

Authors: Saurabh Farkya, Govinda Surampudi

Abstract:

Optical Character Recognition is one of the current major research areas. This paper is focussed on recognition of Devanagari script and its sound generation. This Paper consists of two parts. First, Optical Character Recognition of Devnagari handwritten Script. Second, speech synthesis of the recognized text. This paper shows an implementation of support vector machines for the purpose of Devnagari Script recognition. The Support Vector Machines was trained with Multi Domain features; Transform Domain and Spatial Domain or Structural Domain feature. Transform Domain includes the wavelet feature of the character. Structural Domain consists of Distance Profile feature and Gradient feature. The Segmentation of the text document has been done in 3 levels-Line Segmentation, Word Segmentation, and Character Segmentation. The pre-processing of the characters has been done with the help of various Morphological operations-Otsu's Algorithm, Erosion, Dilation, Filtration and Thinning techniques. The Algorithm was tested on the self-prepared database, a collection of various handwriting. Further, Unicode was used to convert recognized Devnagari text into understandable computer document. The document so obtained is an array of codes which was used to generate digitized text and to synthesize Hindi speech. Phonemes from the self-prepared database were used to generate the speech of the scanned document using concatenation technique.

Keywords: Character Recognition (OCR), Text to Speech (TTS), Support Vector Machines (SVM), Library of Support Vector Machines (LIBSVM)

Procedia PDF Downloads 475

Authors: Sujata S. Kulkarni

Abstract:

Effective and efficient texture feature extraction and classification is an important problem in image understanding and recognition. This paper gives a review on effective texture classification method. The objective of the problem of texture representation is to reduce the amount of raw data presented by the image, while preserving the information needed for the task. Texture analysis is important in many applications of computer image analysis for classification include industrial and biomedical surface inspection, for example for defects and disease, ground classification of satellite or aerial imagery and content-based access to image databases.

Keywords: compressed sensing, feature extraction, image classification, texture analysis

Procedia PDF Downloads 412

Authors: Balpreet Singh Madan

Abstract:

The Middle Eastern countries have preserved their heritage, tradition, and culture in their buildings by incorporating vernacular features of Arab-Islamic Architecture. The harsh sun and arid climate in the Gulf region make their buildings and infrastructure extremely hot and challenging to live in. One such iconic feature of Arab architecture is the Mashrabiya, which has been refined and updated for both functional and aesthetic purposes. This feature helps reduce the impact of solar radiation in buildings and lowers the energy requirements for creating livable conditions. The incorporation of Mashrabiya in modern buildings in the region symbolizes the amalgamation of tradition with innovation and modern technology. These buildings depict Mashrabiya with refinements for its better functional performance and aesthetic appeal to make superior built forms. This paper emphasizes the study of Mashrabiya as a vernacular feature with its adaptability for Energy Conservation and Sustainability, as seen in some of the recent iconic buildings of the Middle East, through a literature review and case studies of renowned buildings.

Keywords: energy efficiency, climate responsive, sustainability, innovation, heritage, vernacular

Procedia PDF Downloads 81

Authors: Zhaoan Wang

Abstract:

Research in predictive maintenance modeling has improved in the recent years to predict failures and needed maintenance with high accuracy, saving cost and improving manufacturing efficiency. However, classic prediction models provide little valuable insight towards the most important features contributing to the failure. By analyzing and quantifying feature importance in predictive maintenance models, cost saving can be optimized based on business goals. First, multiple classifiers are evaluated with cross-validation to predict the multi-class of failures. Second, predictive performance with features provided by different feature selection algorithms are further analyzed. Third, features selected by different algorithms are ranked and combined based on their predictive power. Finally, linear explainer SHAP (SHapley Additive exPlanations) is applied to interpret classifier behavior and provide further insight towards the specific roles of features in both local predictions and global model behavior. The results of the experiments suggest that certain features play dominant roles in predictive models while others have significantly less impact on the overall performance. Moreover, for multi-class prediction of machine failures, the most important features vary with type of machine failures. The results may lead to improved productivity and cost saving by prioritizing sensor deployment, data collection, and data processing of more important features over less importance features.

Keywords: automated supply chain, intelligent manufacturing, predictive maintenance machine learning, feature engineering, model interpretation

Procedia PDF Downloads 113

Authors: Liu Li, Kim Seng Lee, Li Lu

Abstract:

The high-energy-density Li-rich layered materials are promising cathode materials for the next-generation high-performance lithium-ion batteries. They have attracted a lot of attentions due mainly to their high reversible capacity of more than 250 mAh•g-1 at low charge-discharge current. However several drawbacks still hinder their applications, such as voltage decay caused by an undesired phase transformation during cycling and poor rate capability. To conquer these issues, the authors applied F modification methods on the pristine Li1.2Mn0.54Ni0.13Co0.13O2 to enhance its electrochemical performance.

Keywords: Li-ion battery, Li-rich layered cathode material, phase transformation, cycling stability, rate capability

Procedia PDF Downloads 312

Authors: Emil F. Khisamutdinov

Abstract:

Development of functional materials undergoing structural transformations in response to an external stimulus such as environmental changes (pH, temperature, etc.), the presence of particular proteins, or short oligonucleotides are of great interest for a variety of applications ranging from medicine to electronics. The dynamic operations of most nucleic acid (NA) devices, including circuits, nano-machines, and biosensors, rely on networks of NA strand displacement processes in which an external or stimulus strand displaces a target strand from a DNA or RNA duplex. The rate of strand displacement can be greatly increased by the use of “toeholds,” single-stranded regions of the target complex to which the invading strand can bind to initiate the reaction, forming additional base pairs that provide a thermodynamic driving force for transformation. Herein, we developed a highly robust nanoparticle shape transition, sequentially transforming DNA polygons from one shape to another using the toehold-mediated DNA strand displacement technique. The shape transformation was confirmed by agarose gel electrophoresis and atomic force microscopy. Furthermore, we demonstrate that our approach is applicable for RNA shape transformation from triangle to square, which can be detected by fluorescence emission from malachite green binding RNA aptamer. Using gel-shift and fluorescence assays, we demonstrated efficient transformation occurs at isothermal conditions (37°C) that can be implemented within living cells as reporter molecules. This work is intended to provide a simple, cost-effective, and straightforward model for the development of biosensors and regulatory devices in nucleic acid nanotechnology.

Keywords: RNA nanotechnology, bionanotechnology, toehold mediated DNA switch, RNA split fluorogenic aptamers

Procedia PDF Downloads 59

Authors: Y. Pala, M. O. Ertas

Abstract:

In this paper, the general Riccati equation is analytically solved by a new transformation. By the method developed, looking at the transformed equation, whether or not an explicit solution can be obtained is readily determined. Since the present method does not require a proper solution for the general solution, it is especially suitable for equations whose proper solutions cannot be seen at first glance. Since the transformed second order linear equation obtained by the present transformation has the simplest form that it can have, it is immediately seen whether or not the original equation can be solved analytically. The present method is exemplified by several examples.

Keywords: Riccati equation, analytical solution, proper solution, nonlinear

Procedia PDF Downloads 335

Authors: Yixiu Liu, Yunzhou Zhang, Jianning Chi, Hao Chu, Rui Zheng, Libo Sun, Guanghao Chen, Fangtong Zhou

Abstract:

In the last few years, large-scale person re-identification has attracted a lot of attention from video surveillance since it has a potential application prospect in public safety management. However, it is still a challenging job considering the variation in human pose, the changing illumination conditions and the lack of paired samples. Although the accuracy has been significantly improved, the data dependence of the sample training is serious. To tackle this problem, a new strategy is proposed based on bag of visual words (BoVW) model of designing the feature representation which has been widely used in the field of image retrieval. The local features are extracted, and more discriminative feature representation is obtained by cross-view dictionary learning (CDL), then the assignment map is obtained through k-means clustering. Finally, the BoVW histograms are formed which encodes the images with the statistics of the feature classes in the assignment map. Experiments conducted on the CUHK03, Market1501 and MARS datasets show that the proposed method performs favorably against existing approaches.

Keywords: bag of visual words, cross-view dictionary learning, person re-identification, reranking

Procedia PDF Downloads 174

Authors: Fawaz Saad, Hamid Alalwany

Abstract:

Sweden has succeeded to maintain a high level of growth and development and has managed to sustain highly ranked position among the world’s developed countries. In this regard, Swedish universities are playing a vital role in supporting innovation and entrepreneurship at all levels and developing Swedish knowledge economy. This paper is aiming to draw on the experiences of two leading Swedish universities, addressing their transformation approach to create entrepreneurial universities and fulfilling their objectives in the era of knowledge economy. The objectives of the paper include: (1) Introducing the Swedish higher education and its characteristics. (2) Examining the infrastructure elements for innovation and Entrepreneurship at two of the Swedish entrepre-neurial universities. (3) Addressing the key aspects of support systems in the initiatives of both Chalmers and Gothenburg universities to support innovation and advance entrepreneurial practices. The paper will contribute to two discourses: (1) Examining the relationship between support systems for innovation and entrepreneurship and the Universities’ policies and practices. (2) Lessons for University leaders to assist the development and implementation of effective innovation and en-trepreneurship policies and practices.

Keywords: Entrepreneurial University, Chalmers University, Gothenburg University, innovation and entrepreneurship policies, entrepreneurial transformation

Procedia PDF Downloads 492

Authors: Felix Givois, Nicolas R. Gauger, Matthias Kabel

Abstract:

The efficient digital material characterization is of great interest to many fields of application. It consists of the following three steps. First, a 3D reconstruction of 2D scans must be performed. Then, the resulting gray-value image of the material sample is enhanced by image processing methods. Finally, partial differential equations (PDE) are solved on the segmented image, and by averaging the resulting solutions fields, effective properties like stiffness or conductivity can be computed. Due to the high resolution of current CT images, the latter is typically performed with matrix-free solvers. Among them, a solver that uses the explicit formula of the Green-Eshelby operator in Fourier space has been proposed by Moulinec and Suquet. Its algorithmic, most complex part is the Fast Fourier Transformation (FFT). In our talk, we will discuss the potential quantum advantage that can be obtained by replacing the FFT with the Quantum Fourier Transformation (QFT). We will especially show that the data transfer for noisy intermediate-scale quantum (NISQ) devices can be improved by using appropriate boundary conditions for the PDE, which also allows using semi-classical versions of the QFT. In the end, we will compare the results of the QFT-based algorithm for simple geometries with the results of the FFT-based homogenization method.

Keywords: most likelihood amplitude estimation (MLQAE), numerical homogenization, quantum Fourier transformation (QFT), NISQ devises

Procedia PDF Downloads 57

Authors: Yanjun Chen, Chuanping Hu, Jie Shao, Lin Mei, Chongyang Zhang

Abstract:

Traditional image classification is based on encoding scheme (e.g. Fisher Vector, Vector of Locally Aggregated Descriptor) with low-level image features (e.g. SIFT, HoG). Compared to these low-level local features, deep convolutional features obtained at the mid-level layer of convolutional neural networks (CNN) have richer information but lack of geometric invariance. For scene classification, there are scattered objects with different size, category, layout, number and so on. It is crucial to find the distinctive objects in scene as well as their co-occurrence relationship. In this paper, we propose a method to take advantage of both deep convolutional features and the traditional encoding scheme while taking object-centric and scene-centric information into consideration. First, to exploit the object-centric and scene-centric information, two CNNs that trained on ImageNet and Places dataset separately are used as the pre-trained models to extract deep convolutional features at multiple scales. This produces dense local activations. By analyzing the performance of different CNNs at multiple scales, it is found that each CNN works better in different scale ranges. A scale-wise CNN adaption is reasonable since objects in scene are at its own specific scale. Second, a fisher kernel is applied to aggregate a global representation at each scale and then to merge into a single vector by using a post-processing method called scale-wise normalization. The essence of Fisher Vector lies on the accumulation of the first and second order differences. Hence, the scale-wise normalization followed by average pooling would balance the influence of each scale since different amount of features are extracted. Third, the Fisher vector representation based on the deep convolutional features is followed by a linear Supported Vector Machine, which is a simple yet efficient way to classify the scene categories. Experimental results show that the scale-specific feature extraction and normalization with CNNs trained on object-centric and scene-centric datasets can boost the results from 74.03% up to 79.43% on MIT Indoor67 when only two scales are used (compared to results at single scale). The result is comparable to state-of-art performance which proves that the representation can be applied to other visual recognition tasks.

Keywords: deep convolutional features, Fisher Vector, multiple scales, scale-specific normalization

Procedia PDF Downloads 313

Authors: Alemu Moges Belay, Torgeir Welo, Jan Ola Strandhagen

Abstract:

Lean has long been linked with manufacturing, but its application claimed also by other functions such as product development and services. However, there is a challenge on understanding and defining lean in each function context. This paper aims to investigate the literature that focus mainly on PD process improvement, obtain better understanding and redefine LPD in systematic way. In addition to that, the paper attempts to summarize various proposed transformation strategies, definitions, identifying features of manufacturing and product development that would help to redefining lean in product development context. Finally we redefine LPD in organized way that encompasses different steps such as stage gate, communication and information, events, learning, innovation, knowledge and value creation.

Keywords: lean, lean manufacturing, lean product development, transformation, strategies

Procedia PDF Downloads 456

Authors: Mahdieh Khalilinezhad, Silvana Dellepiane, Gianni Vernazza

Abstract:

The aim of the present work is to build a model based on tissue characterization that is able to discriminate pathological and non-pathological regions from three-phasic CT images. Based on feature selection in different phases, in this research, we design a neural network system that has optimal neuron number in a hidden layer. Our approach consists of three steps: feature selection, feature reduction, and classification. For each ROI, 6 distinct set of texture features are extracted such as first order histogram parameters, absolute gradient, run-length matrix, co-occurrence matrix, autoregressive model, and wavelet, for a total of 270 texture features. We show that with the injection of liquid and the analysis of more phases the high relevant features in each region changed. Our results show that for detecting HCC tumor phase3 is the best one in most of the features that we apply to the classification algorithm. The percentage of detection between these two classes according to our method, relates to first order histogram parameters with the accuracy of 85% in phase 1, 95% phase 2, and 95% in phase 3.

Keywords: multi-phasic liver images, texture analysis, neural network, hidden layer

Procedia PDF Downloads 246

Authors: Joshua van Staden, Dane Brown, Karen Bradshaw

Abstract:

Gun violence is a significant problem in modern-day society. Early detection of carried handguns through closed-circuit television (CCTV) can aid in preventing potential gun violence. However, CCTV operators have a limited attention span. Machine learning approaches to automating the detection of dangerous gun carriers provide a way to aid CCTV operators in identifying these individuals. This study provides insight into the relationship between human key points extracted using human pose estimation (HPE) and their intention to fire a weapon. We examine the feature importance of each keypoint and their correlations. We use principal component analysis (PCA) to reduce the feature space and optimize detection. Finally, we run a set of classifiers to determine what form of classifier performs well on this data. We find that hips, shoulders, and knees tend to be crucial aspects of the human pose when making these predictions. Furthermore, the horizontal position plays a larger role than the vertical position. Of the 66 key points, nine principal components could be used to make nonlinear classifications with 86% accuracy. Furthermore, linear classifications could be done with 85% accuracy, showing that there is a degree of linearity in the data.

Keywords: feature engineering, human pose, machine learning, security

Procedia PDF Downloads 80

Authors: Maurice Ntahobari, Levin Kuhlmann, Mario Boley, Zhinoos Razavi Hesabi

Abstract:

For machine learning based epileptic seizure prediction, it is important for the model to be implemented in small implantable or wearable devices that can be used to monitor epilepsy patients; however, current state-of-the-art methods are complex and computationally intensive. We use Shapley Additive Explanation (SHAP) to find relevant intracranial electroencephalogram (iEEG) features and improve the computational efficiency of a state-of-the-art seizure prediction method based on the extra trees classifier while maintaining prediction performance. Results for a small contest dataset and a much larger dataset with continuous recordings of up to 3 years per patient from 15 patients yield better than chance prediction performance (p < 0.004). Moreover, while the performance of the SHAP-based model is comparable to that of the benchmark, the overall training and prediction time of the model has been reduced by a factor of 1.83. It can also be noted that the feature called zero crossing value is the best EEG feature for seizure prediction. These results suggest state-of-the-art seizure prediction performance can be achieved using efficient methods based on optimal feature selection.

Keywords: machine learning, seizure prediction, extra tree classifier, SHAP, epilepsy

Procedia PDF Downloads 93

Authors: Fabiola M. Villalobos-Castaldi, Nicolás C. Kemper, Esther Rojas-Krugger, Laura G. Ramírez-Sánchez

Abstract:

This paper presents the classification results about exploring the combination of a motion based approach with a local appearance method to describe the facial motion caused by the muscle contractions and expansions that are presented in facial expressions. The proposed feature extraction method take advantage of the knowledge related to which parts of the face reflects the highest deformations, so we selected 4 specific facial regions at which the appearance descriptor were applied. The most common used approaches for feature extraction are the holistic and the local strategies. In this work we present the results of using a local appearance approach estimating the correlation coefficient to the 4 corresponding landmark-localized facial templates of the expression face related to the neutral face. The results let us to probe how the proposed motion estimation scheme based on the local appearance correlation computation can simply and intuitively measure the motion parameters for some of the most relevant facial regions and how these parameters can be used to recognize facial expressions automatically.

Keywords: facial expression recognition system, feature extraction, local-appearance method, motion-based approach

Procedia PDF Downloads 395

Authors: Soyon Kim, Edward Kim

Abstract:

In order to retrieve information from the massive stream of songs in the music industry, music search by title, lyrics, artist, mood, and genre has become more important. Despite the subjectivity and controversy over the definition of music genres across different nations and cultures, automatic genre classification systems that facilitate the process of music categorization have been developed. Manual genre selection by music producers is being provided as statistical data for designing automatic genre classification systems. In this paper, an automatic music genre classification system utilizing non-negative matrix factorization (NMF) is proposed. Short-term characteristics of the music signal can be captured based on the timbre features such as mel-frequency cepstral coefficient (MFCC), decorrelated filter bank (DFB), octave-based spectral contrast (OSC), and octave band sum (OBS). Long-term time-varying characteristics of the music signal can be summarized with (1) the statistical features such as mean, variance, minimum, and maximum of the timbre features and (2) the modulation spectrum features such as spectral flatness measure, spectral crest measure, spectral peak, spectral valley, and spectral contrast of the timbre features. Not only these conventional basic long-term feature vectors, but also NMF based feature vectors are proposed to be used together for genre classification. In the training stage, NMF basis vectors were extracted for each genre class. The NMF features were calculated in the log spectral magnitude domain (NMF-LSM) as well as in the basic feature vector domain (NMF-BFV). For NMF-LSM, an entire full band spectrum was used. However, for NMF-BFV, only low band spectrum was used since high frequency modulation spectrum of the basic feature vectors did not contain important information for genre classification. In the test stage, using the set of pre-trained NMF basis vectors, the genre classification system extracted the NMF weighting values of each genre as the NMF feature vectors. A support vector machine (SVM) was used as a classifier. The GTZAN multi-genre music database was used for training and testing. It is composed of 10 genres and 100 songs for each genre. To increase the reliability of the experiments, 10-fold cross validation was used. For a given input song, an extracted NMF-LSM feature vector was composed of 10 weighting values that corresponded to the classification probabilities for 10 genres. An NMF-BFV feature vector also had a dimensionality of 10. Combined with the basic long-term features such as statistical features and modulation spectrum features, the NMF features provided the increased accuracy with a slight increase in feature dimensionality. The conventional basic features by themselves yielded 84.0% accuracy, but the basic features with NMF-LSM and NMF-BFV provided 85.1% and 84.2% accuracy, respectively. The basic features required dimensionality of 460, but NMF-LSM and NMF-BFV required dimensionalities of 10 and 10, respectively. Combining the basic features, NMF-LSM and NMF-BFV together with the SVM with a radial basis function (RBF) kernel produced the significantly higher classification accuracy of 88.3% with a feature dimensionality of 480.

Keywords: mel-frequency cepstral coefficient (MFCC), music genre classification, non-negative matrix factorization (NMF), support vector machine (SVM)

Procedia PDF Downloads 278

Authors: Linyu Wang, Liguo Qiao, Jianhong Xiang, Hao Xu

Abstract:

In semantic communication, the existing joint Source Channel coding (JSCC) wireless communication system without pilot has unstable transmission performance and can not effectively capture the global information and location information of images. In this paper, a pilot-free image transmission system of joint source channel based on multi-level semantic information (Multi-level JSCC) is proposed. The transmitter of the system is composed of two networks. The feature extraction network is used to extract the high-level semantic features of the image, compress the information transmitted by the image, and improve the bandwidth utilization. Feature retention network is used to preserve low-level semantic features and image details to improve communication quality. The receiver also is composed of two networks. The received high-level semantic features are fused with the low-level semantic features after feature enhancement network in the same dimension, and then the image dimension is restored through feature recovery network, and the image location information is effectively used for image reconstruction. This paper verifies that the proposed multi-level JSCC algorithm can effectively transmit and recover image information in both AWGN channel and Rayleigh fading channel, and the peak signal-to-noise ratio (PSNR) is improved by 1~2dB compared with other algorithms under the same simulation conditions.

Keywords: deep learning, JSCC, pilot-free picture transmission, multilevel semantic information, robustness

Procedia PDF Downloads 100