World Academy of Science, Engineering and Technology

Authors: Saumya Srivastava, Rina Maiti

Abstract:

In order to achieve a better performance of vehicle detection in a complex environment, we present an efficient approach for a multi-vehicle detection system using an adaptive sliding window technique. For a given frame, image segmentation is carried out to establish the region of interest. Gradient computation followed by thresholding, denoising, and morphological operations is performed to extract the binary search image. Near-region field and far-region field are defined to generate hypotheses using the adaptive sliding window technique on the resultant binary search image. For each vehicle candidate, features are extracted using a histogram of oriented gradients, and a pre-trained support vector machine is applied for hypothesis verification. Later, the Kalman filter is used for tracking the vanishing point. The experimental results show that the method is robust and effective on various roads and driving scenarios. The algorithm was tested on highways and urban roads in India.

Keywords: gradient, vehicle detection, histograms of oriented gradients, support vector machine

Procedia PDF Downloads 87

Authors: Johanna Perdomo, Riki Lamont, Edmund Pickering, Naomi C. Paxton, Maria A. Woodruff

Abstract:

Background: Tissue Engineering (TE) has combined advanced materials, such as biomaterials, to create affordable scaffolds and dynamic systems to generate stimulation of seeded cells on these scaffolds, improving and maintaining the cellular growth process in a cell culture. However, Few TE skin products have been clinically translated, and more research is required to produce highly biomimetic skin substitutes that mimic the native elasticity of skin in a controlled manner. Therefore, this work will be focused on the fabrication of a novel mechanical system to enhance the TE treatment approaches for the reparation of damaged tissue skin. Aims: To archive this, a soft robotic device will be created to emulate different deformation of skin stress. The design of this soft robot will allow the attachment of scaffolds, which will then be mechanically actuated. This will provide a novel and highly adaptable platform for dynamic cell culture. Methods: Novel, low-cost soft robot is fabricated via 3D printed moulds and silicone. A low cost, electro-mechanical device was constructed to actuate the soft robot through the controlled combination of positive and negative air pressure to control the different state of movements. Mechanical tests were conducted to assess the performance and calibration of each electronic component. Similarly, pressure-displacement test was performed on scaffolds, which were attached to the soft robot, applying various mechanical loading regimes. Lastly, digital image correlation test was performed to obtain strain distributions over the soft robot’s surface. Results: The control system can control and stabilise positive pressure changes for long hours. Similarly, pressure-displacement test demonstrated that scaffolds with 5µm of diameter and wavy geometry can displace at 100%, applying a maximum pressure of 1.5 PSI. Lastly, during the inflation state, the displacement of silicone was measured using DIC method, and this showed a parameter of 4.78 mm and strain of 0.0652. Discussion And Conclusion: The developed soft robot system provides a novel and low-cost platform for the dynamic actuation of tissue scaffolds with a target towards dynamic cell culture.

Keywords: soft robot, tissue engineering, mechanical stimulation, dynamic cell culture, bioreactor

Procedia PDF Downloads 63

Authors: Asma Lamin, Anna H. Kaksonen, Ivan Cole, Paul White, Xiao-Bo Chen

Abstract:

Microbiologically influenced corrosion (MIC) is a process in which microorganisms initiate, facilitate, or accelerate the electrochemical corrosion reactions of metallic components. Several reports documented that MIC accounts for about 20 to 40 % of the total cost of corrosion. Biofilm formation due to the presence of microorganisms on the surface of metal components is known to play a vital role in MIC, which can lead to severe consequences in various environmental and industrial settings. Quorum sensing (QS) system plays a major role in regulating biofilm formation and control the expression of some microbial enzymes. QS is a communication mechanism between microorganisms that involves the regulation of gene expression as a response to the microbial cell density within an environment. This process is employed by both Gram-positive and Gram-negative bacteria to regulate different physiological functions. QS involves production, detection, and responses to signalling chemicals, known as auto-inducers. QS controls specific processes important for the microbial community, such as biofilm formation, virulence factor expression, production of secondary metabolites and stress adaptation mechanisms. The use of QS inhibitors (QSIs) has been proposed as a possible solution to biofilm related challenges in many different applications. Although QSIs have demonstrated some strength in tackling biofouling, QSI-based strategies to control microbially influenced corrosion have not been thoroughly investigated. As such, our research aims to target the QS mechanisms as a strategy for mitigating MIC on metal surfaces in engineered systems.

Keywords: quorum sensing, quorum quenching, biofilm, biocorrosion

Procedia PDF Downloads 59

Authors: Ashraf Nawaz Khan, R. Alagirusamy, Apurba Das, Puneet Mahajan

Abstract:

The thermoplastic-based fibre composites are acquiring a market sector of conventional as well as thermoset composites. However, replacing the thermoset with a thermoplastic composite has never been an easy task. The inherent high viscosity of thermoplastic resin reveals poor interface properties. In this work, a homo-polymer towpreg is produced through an electrostatic powder spray coating methodology. The produced flexible towpreg offers a low melt-flow distance during the consolidation of the laminate. The reduced melt-flow distance demonstrates a homogeneous fibre/matrix distribution (and low void content) on consolidation. The composite laminate has been fabricated with two manufacturing techniques such as conventional film stack (FS) and powder-coated (PC) technique. This helps in understanding the distinct response of produced laminates on applying load since the laminates produced through the two techniques are comprised of the same constituent fibre and matrix (constant fibre volume fraction). The changed behaviour is observed mainly due to the different fibre/matrix configurations within the laminate. The interface adhesion influences the load transfer between the fibre and matrix. Therefore, it influences the elastic, plastic, and failure patterns of the laminates. Moreover, the effect of preform geometries (plain weave and satin weave structure) are also studied for corresponding composite laminates in terms of various mechanical properties. The fracture analysis is carried out to study the effect of resin at the interlacement points through micro-CT analysis. The PC laminate reveals a considerably small matrix-rich and deficient zone in comparison to the FS laminate. The different load tensile, shear, fracture toughness, and drop weight impact test) is applied to the laminates, and corresponding damage behaviour is analysed in the successive stage of failure. The PC composite has shown superior mechanical properties in comparison to the FS composite. The damage that occurs in the laminate is captured through the SEM analysis to identify the prominent mode of failure, such as matrix cracking, fibre breakage, delamination, debonding, and other phenomena.

Keywords: composite, damage, fibre, manufacturing

Procedia PDF Downloads 108

Authors: Janusz Bobulski, Kamila Pasternak

Abstract:

Despite many years of effort and research, the problem of waste management is still current. So far, no fully effective waste management system has been developed. Many programs and projects improve statistics on the percentage of waste recycled every year. In these efforts, it is worth using modern Computer Vision techniques supported by artificial intelligence. In the article, we present a method of identifying plastic waste based on the asymmetry analysis of the histogram of the image containing the waste. The method is simple but effective (94%), which allows it to be implemented on devices with low computing power, in particular on microcomputers. Such de-vices will be used both at home and in waste sorting plants.

Keywords: waste management, environmental protection, image processing, computer vision

Procedia PDF Downloads 76

Authors: Magdalena Wilk-Kozubek, Magdalena Rowińska, Krzysztof Rola, Joanna Cybińska

Abstract:

Leuco dye-based thermochromic systems are classified as intelligent materials because they exhibit thermally induced color changes. Thanks to this feature, they are mainly used as temperature sensors in many industrial sectors. For example, placing a thermochromic material on a chemical reactor may warn about exceeding the maximum permitted temperature for a chemical process. Usually two components, a color former and a developer are needed to produce a system with irreversible color change. The color former is an electron donating (proton accepting) compound such as fluoran leuco dye. The developer is an electron accepting (proton donating) compound such as organic carboxylic acid. When the developer melts, the color former - developer complex is created and the termochromic system becomes colored. Typically, the melting point of the applied developer determines the temperature at which the color change occurs. When the lactone ring of the color former is closed, then the dye is in its colorless state. The ring opening, induced by the addition of a proton, causes the dye to turn into its colored state. Since the color former and the developer are often solid, they can be incorporated into polymer films to facilitate their practical use in industry. The objective of this research was to fabricate a leuco dye-based termochromic system that will irreversibly change color after reaching the temperature of 100°C. For this purpose, benzofluoran leuco dye (as color former) and phenoxyacetic acid (as developer with a melting point of 100°C) were introduced into the polymer films during the drop casting process. The film preparation process was optimized in order to obtain thin films with appropriate properties such as transparency, flexibility and homogeneity. Among the optimized factors were the concentration of benzofluoran leuco dye and phenoxyacetic acid, the type, average molecular weight and concentration of the polymer, and the type and concentration of the surfactant. The selected films, containing benzofluoran leuco dye and phenoxyacetic acid, were combined by mild heat treatment. Structural characterization of single and combined films was carried out by FTIR spectroscopy, morphological analysis was performed by optical microscopy and SEM, phase transitions were examined by DSC, color changes were investigated by digital photography and UV-Vis spectroscopy, while emission changes were studied by photoluminescence spectroscopy. The resulting thermochromic system is colorless at room temperature, but after reaching 100°C the developer melts and it turns irreversibly pink. Therefore, it could be used as an additional sensor to warn against boiling of water in power plants using water cooling. Currently used electronic temperature indicators are prone to faults and unwanted third-party actions. The sensor constructed in this work is transparent, thanks to which it can be unnoticed by an outsider and constitute a reliable reference for the person responsible for the apparatus.

Keywords: color developer, leuco dye, thin film, thermochromism

Procedia PDF Downloads 70

Authors: Ashish James, Anusha James

Abstract:

The impressive predictive performance of deep learning techniques on a wide range of tasks has led to its widespread use. Estimating the confidence of these predictions is paramount for improving the safety and reliability of such systems. However, the uncertainty estimates provided by neural networks (NNs) tend to be overconfident and unreasonable. Ensemble of NNs typically produce good predictions but uncertainty estimates tend to be inconsistent. Inspired by these, this paper presents a framework that can quantitatively estimate the uncertainties by leveraging the advances in transfer learning through slight modification to the existing training pipelines. This promising algorithm is developed with an intention of deployment in real world problems which already boast a good predictive performance by reusing those pretrained models. The idea is to capture the behavior of the trained NNs for the base task by augmenting it with the uncertainty estimates from a supplementary network. A series of experiments with known and unknown distributions show that the proposed approach produces well calibrated uncertainty estimates with high quality predictions.

Keywords: uncertainty estimation, neural networks, transfer learning, regression

Procedia PDF Downloads 95

Authors: Meshandren Naidoo, Amy Gooden

Abstract:

Implantable brain-computer interface (BCI) technologies have developed to the point where brain-computer communication is possible. This has great potential in the medical field, as it allows persons who have lost capacities. However, ethicists and regulators call for a strict approach to these technologies due to the impact on personhood. This research demonstrates that the personhood debate is more nuanced and that where an African approach to personhood is used, it may produce results more favorable to the development and use of this technology.

Keywords: artificial intelligence, law, neuroscience, ethics

Procedia PDF Downloads 88

Authors: Blessing Ojeme, Frederick Quinn, Russell Karls, Shannon Quinn

Abstract:

The detection and segmentation of mitochondria from fluorescence microscopy are crucial for understanding the complex structure of the nervous system. However, the constant fission and fusion of mitochondria and image distortion in the background make the task of detection and segmentation challenging. In the literature, a number of open-source software tools and artificial intelligence (AI) methods have been described for analyzing mitochondrial images, achieving remarkable classification and quantitation results. However, the availability of combined expertise in the medical field and AI required to utilize these tools poses a challenge to its full adoption and use in clinical settings. Motivated by the advantages of automated methods in terms of good performance, minimum detection time, ease of implementation, and cross-platform compatibility, this study proposes a fully automated framework for the detection and segmentation of mitochondria using both image shape information and descriptive statistics. Using the low-cost, open-source python and openCV library, the algorithms are implemented in three stages: pre-processing, image binarization, and coarse-to-fine segmentation. The proposed model is validated using the mitochondrial fluorescence dataset. Ground truth labels generated using a Lab kit were also used to evaluate the performance of our detection and segmentation model. The study produces good detection and segmentation results and reports the challenges encountered during the image analysis of mitochondrial morphology from the fluorescence mitochondrial dataset. A discussion on the methods and future perspectives of fully automated frameworks conclude the paper.

Keywords: 2D, binarization, CLAHE, detection, fluorescence microscopy, mitochondria, segmentation

Procedia PDF Downloads 332

Authors: Mario Robinson, Moe Rabea

Abstract:

This research examines the effects of heat treatment processes on the mechanical properties and corrosion resistanceof1045 and 4140 Steel Alloysfor industrial applications. Heat treatment processes of full annealing, normalizing, quenching, and tempering are carried out on the alloy samples. The mechanical and corrosion resistance tests of the heat treated samples are carried out, and the results obtained are related to their SEMmorphologies analysis. The results show that the heat treatment processes have an effect on the tensile strength, impact, and a significant effect on the corrosion resistance of the alloy samples. With respect to the strain characteristics, significant improvement in the ductility of the samples is recorded in the full annealing and alloy tempered samples. Thus, for application requiring strength and ductility, such as in aerospace industries, this tempered heat treated alloy could be used. In addition, the quenched sample shows a significant improvement in hardness.

Keywords: heat treatment, corrosion resistance, steel, industrial appilcations

Procedia PDF Downloads 143

Authors: Anamitra Ghosh, Neeladri Paul

Abstract:

The present work aims at investigating material combinations and thereby improvising an optimized design of liner-mother plate arrangement and that of the stone box, such that it has low cost, high weldability, sufficiently capable of withstanding the increased amount of corrosive shear and bending loads, and having reduced thermal expansion coefficient at temperatures close to 1000 degrees Celsius. All the above factors have been preliminarily examined using a computational approach via ANSYS Thermo-Structural Computation, a commercial software that uses the Finite Element Method to analyze the response of simulated design specimens of liner-mother plate arrangement and the stone box, to varied bending, shear, and thermal loads as well as to determine the temperature gradients developed across various surfaces of the designs. Finally, the optimized structural designs of the liner-mother plate arrangement and that of the stone box with improved material and better structural and thermal properties are selected via trial-and-error method. The final improvised design is therefore considered to enhance the overall life and reliability of a Direct Charge Transfer Chute that transfers and segregates the hot sinter onto the cooler in a sinter plant.

Keywords: shear, bending, thermal, sinter, simulated, optimized, charge, transfer, chute, expansion, computational, corrosive, stone box, liner, mother plate, arrangement, material

Procedia PDF Downloads 76

Authors: A. Shoiynbek, K. Kozhakhmet, P. Menezes, D. Kuanyshbay, D. Bayazitov

Abstract:

Speech emotion recognition has received increasing research interest all through current years. There was used emotional speech that was collected under controlled conditions in most research work. Actors imitating and artificially producing emotions in front of a microphone noted those records. There are four issues related to that approach, namely, (1) emotions are not natural, and it means that machines are learning to recognize fake emotions. (2) Emotions are very limited by quantity and poor in their variety of speaking. (3) There is language dependency on SER. (4) Consequently, each time when researchers want to start work with SER, they need to find a good emotional database on their language. In this paper, we propose the approach to create an automatic tool for speech emotion extraction based on facial emotion recognition and describe the sequence of actions of the proposed approach. One of the first objectives of the sequence of actions is a speech detection issue. The paper gives a detailed description of the speech detection model based on a fully connected deep neural network for Kazakh and Russian languages. Despite the high results in speech detection for Kazakh and Russian, the described process is suitable for any language. To illustrate the working capacity of the developed model, we have performed an analysis of speech detection and extraction from real tasks.

Keywords: deep neural networks, speech detection, speech emotion recognition, Mel-frequency cepstrum coefficients, collecting speech emotion corpus, collecting speech emotion dataset, Kazakh speech dataset

Procedia PDF Downloads 66

Authors: Milad Abbasi

Abstract:

Carbon fiber-reinforced polymer (CFRP) composite structures are increasingly being utilized in the automotive industry due to their lightweight and specific energy absorption capabilities. Although it is impossible to predict composite mechanical properties directly using theoretical methods, various research has been conducted so far in the literature for accurate simulation of CFRP structures' energy-absorbing behavior. In this research, axial compression experiments were carried out on hand lay-up unidirectional CFRP composite tubes. The fabrication method allowed the authors to extract the material properties of the CFRPs using ASTM D3039, D3410, and D3518 standards. A neural network machine learning algorithm was then utilized to build a robust prediction model to forecast the axial compressive properties of CFRP tubes while reducing high-cost experimental efforts. The predicted results have been compared with the experimental outcomes in terms of load-carrying capacity and energy absorption capability. The results showed high accuracy and precision in the prediction of the energy-absorption capacity of the CFRP tubes. This research also demonstrates the effectiveness and challenges of machine learning techniques in the robust simulation of composites' energy-absorption behavior. Interestingly, the proposed method considerably condensed numerical and experimental efforts in the simulation and calibration of CFRP composite tubes subjected to compressive loading.

Keywords: CFRP composite tubes, energy absorption, crushing behavior, machine learning, neural network

Procedia PDF Downloads 103

Authors: Wu Peng, Anders Liljerehn, Martin Magnevall

Abstract:

In metal cutting, tool wear gradually changes the micro geometry of the cutting edge. Today there is a significant gap in understanding the impact these geometrical changes have on the cutting forces which governs tool deflection and heat generation in the cutting zone. Accurate models and understanding of the interaction between the work piece and cutting tool leads to improved accuracy in simulation of the cutting process. These simulations are useful in several application areas, e.g., optimization of insert geometry and machine tool monitoring. This study aims to develop an extended Kienzle force model to account for the effect of rake angle variations and tool flank wear have on the cutting forces. In this paper, the starting point sets from cutting force measurements using orthogonal turning tests of pre-machined flanches with well-defined width, using triangular coated inserts to assure orthogonal condition. The cutting forces have been measured by dynamometer with a set of three different rake angles, and wear progression have been monitored during machining by an optical measuring collaborative robot. The method utilizes the measured cutting forces with the inserts flank wear progression to extend the mechanistic cutting forces model with flank wear as an input parameter. The adapted cutting forces model is validated in a turning process with commercial cutting tools. This adapted cutting forces model shows the significant capability of prediction of cutting forces accounting for tools flank wear and different-rake-angle cutting tool inserts. The result of this study suggests that the nonlinear effect of tools flank wear and interaction between the work piece and the cutting tool can be considered by the developed cutting forces model.

Keywords: cutting force, kienzle model, predictive model, tool flank wear

Procedia PDF Downloads 71

Authors: Wesley Teskey, Vedran Glavas, Julian Wegener

Abstract:

Generating battery cathode microstructures is an important area of research, given the proliferation of the use of automotive batteries. Currently, finite element analysis (FEA) is often used for simulations of battery cathode microstructures before physical batteries can be manufactured and tested to verify the simulation results. Unfortunately, a key drawback of using FEA is that this method of simulation is very slow in terms of computational runtime. Generative AI offers the key advantage of speed when compared to FEA, and because of this, generative AI is capable of evaluating very large numbers of candidate microstructures. Given AI generated candidate microstructures, a subset of the promising microstructures can be selected for further validation using FEA. Leveraging the speed advantage of AI allows for a better final microstructural selection because high speed allows for the evaluation of many more candidate microstructures. For the approach presented, battery cathode 3D candidate microstructures are generated using Gaussian Mixture Models (GMMs) and pix2pix. This approach first uses GMMs to generate a population of spheres (representing the “active material” of the cathode). Once spheres have been sampled from the GMM, they are placed within a microstructure. Subsequently, the pix2pix sweeps over the 3D microstructure (iteratively) slice by slice and adds details to the microstructure to determine what portions of the microstructure will become electrolyte and what part of the microstructure will become binder. In this manner, each subsequent slice of the microstructure is evaluated using pix2pix, where the inputs into pix2pix are the previously processed layers of the microstructure. By feeding into pix2pix previously fully processed layers of the microstructure, pix2pix can be used to ensure candidate microstructures represent a realistic physical reality. More specifically, in order for the microstructure to represent a realistic physical reality, the locations of electrolyte and binder in each layer of the microstructure must reasonably match the locations of electrolyte and binder in previous layers to ensure geometric continuity. Using the above outlined approach, a 10x to 100x speed increase was possible when generating candidate microstructures using AI when compared to using a FEA only approach for this task. A key metric for evaluating microstructures was the battery specific power value that the microstructures would be able to produce. The best generative AI result obtained was a 12% increase in specific power for a candidate microstructure when compared to what a FEA only approach was capable of producing. This 12% increase in specific power was verified by FEA simulation.

Keywords: finite element analysis, gaussian mixture models, generative design, Pix2Pix, structural design

Procedia PDF Downloads 73

Authors: Marianna Maiaru, Gregory M. Odegard

Abstract:

During the processing of high-performance thermoset polymer matrix composites, chemical reactions occur during elevated pressure and temperature cycles, causing the constituent monomers to crosslink and form a molecular network that gradually can sustain stress. As the crosslinking process progresses, the material naturally experiences a gradual shrinkage due to the increase in covalent bonds in the network. Once the cured composite completes the cure cycle and is brought to room temperature, the thermal expansion mismatch of the fibers and matrix cause additional residual stresses to form. These compounded residual stresses can compromise the reliability of the composite material and affect the composite strength. Composite process modeling is greatly complicated by the multiscale nature of the composite architecture. At the molecular level, the degree of cure controls the local shrinkage and thermal-mechanical properties of the thermoset. At the microscopic level, the local fiber architecture and packing affect the magnitudes and locations of residual stress concentrations. At the macroscopic level, the layup sequence controls the nature of crack initiation and propagation due to residual stresses. The goal of this research is use molecular dynamics (MD) and finite element analysis (FEA) to predict the residual stresses in composite laminates and the corresponding effect on composite failure. MD is used to predict the polymer shrinkage and thermomechanical properties as a function of degree of cure. This information is used as input into FEA to predict the residual stresses on the microscopic level resulting from the complete cure process. Virtual testing is subsequently conducted to predict strength allowables. Experimental characterization is used to validate the modeling.

Keywords: molecular dynamics, finite element analysis, processing modeling, multiscale modeling

Procedia PDF Downloads 64

Authors: Huayuan Lu, Chunfang Yang, Ma Zhu, Baojun Qi, Yaqiong Qiao, Jiangqian Xu

Abstract:

Image geolocalization has great application prospects in fields such as autonomous driving and virtual/augmented reality. In practical application scenarios, the size of the image to be located is not fixed; it is impractical to train different networks for all possible sizes. When its size does not match the size of the input of the descriptor extraction model, existing image geolocalization methods usually directly scale or crop the image in some common ways. This will result in the loss of some information important to the geolocalization task, thus affecting the performance of the image geolocalization method. For example, excessive down-sampling can lead to blurred building contour, and inappropriate cropping can lead to the loss of key semantic elements, resulting in incorrect geolocation results. To address this problem, this paper designs a learnable image resizer and proposes an arbitrary-sized image geolocation method. (1) The designed learnable image resizer employs the self-attention mechanism to enhance the geometric features of the resized image. Firstly, it applies bilinear interpolation to the input image and its feature maps to obtain the initial resized image and the resized feature maps. Then, SKNet (selective kernel net) is used to approximate the best receptive field, thus keeping the geometric shapes as the original image. And SENet (squeeze and extraction net) is used to automatically select the feature maps with strong contour information, enhancing the geometric features. Finally, the enhanced geometric features are fused with the initial resized image, to obtain the final resized images. (2) The proposed image geolocalization method embeds the above image resizer as a fronting layer of the descriptor extraction network. It not only enables the network to be compatible with arbitrary-sized input images but also enhances the geometric features that are crucial to the image geolocalization task. Moreover, the triplet attention mechanism is added after the first convolutional layer of the backbone network to optimize the utilization of geometric elements extracted by the first convolutional layer. Finally, the local features extracted by the backbone network are aggregated to form image descriptors for image geolocalization. The proposed method was evaluated on several mainstream datasets, such as Pittsburgh30K, Tokyo24/7, and Places365. The results show that the proposed method has excellent size compatibility and compares favorably to recently mainstream geolocalization methods.

Keywords: image geolocalization, self-attention mechanism, image resizer, geometric feature

Procedia PDF Downloads 171

Authors: Naman Solanki, Souraj Mondal

Abstract:

With the world moving towards online communication, the video datastore has exploded in the past few years. Consequently, it has become crucial to analyse participant’s engagement levels in online communication videos. Engagement prediction of people in videos can be useful in many domains, like education, client meetings, dating, etc. Video-level or frame-level prediction of engagement for a user involves the development of robust models that can capture facial micro-emotions efficiently. For the development of an engagement prediction model, it is necessary to have a widely-accepted standard dataset for engagement analysis. DAiSEE is one of the datasets which consist of in-the-wild data and has a gold standard annotation for engagement prediction. Earlier research done using the DAiSEE dataset involved training and testing standard models like CNN-based models, but the results were not satisfactory according to industry standards. In this paper, a multi-level classification approach has been introduced to create a more robust model for engagement analysis using the DAiSEE dataset. This approach has recorded testing accuracies of 0.638, 0.7728, 0.8195, and 0.866 for predicting boredom level, engagement level, confusion level, and frustration level, respectively.

Keywords: computer vision, engagement prediction, deep learning, multi-level classification

Procedia PDF Downloads 85

Authors: They Shao Peng

Abstract:

Machine vision system for inspection is an automated technology that is normally utilized to analyze items on the production line for quality control purposes, it also can be known as an automated visual inspection (AVI) system. By applying automated visual inspection, the existence of items, defects, contaminants, flaws, and other irregularities in manufactured products can be easily detected in a short time and accurately. However, AVI systems are still inflexible and expensive due to their uniqueness for a specific task and consuming a lot of set-up time and space. With the rapid development of mobile devices, smartphones can be an alternative device for the visual system to solve the existing problems of AVI. Since the smartphone-based AVI system is still at a nascent stage, this led to the motivation to investigate the smartphone-based AVI system. This study is aimed to provide a low-cost AVI system with high efficiency and flexibility. In this project, the object detection models, which are You Only Look Once (YOLO) model and Single Shot MultiBox Detector (SSD) model, are trained, evaluated, and integrated with the smartphone and webcam devices. The performance of the smartphone-based AVI is compared with the webcam-based AVI according to the precision and inference time in this study. Additionally, a mobile application is developed which allows users to implement real-time object detection and object detection from image storage.

Keywords: automated visual inspection, deep learning, machine vision, mobile application

Procedia PDF Downloads 90

Authors: Aleksei Samarin, Artyom Nazarenko, Valentin Malykh

Abstract:

Image enhancement is becoming more and more popular, especially on mobile devices. Nowadays, it is a common approach to enhance an image using a convolutional neural network (CNN). Such a network should be of significant size; otherwise, a possibility for the artifacts to occur is overgrowing. The existing large CNNs are computationally expensive, which could be crucial for mobile devices. Another important flaw of such models is they are poorly interpretable. There is another approach to image enhancement, namely, the usage of predefined filters in combination with the prediction of their applicability. We present an approach following this paradigm, which outperforms both existing CNN-based and filter-based approaches in the image enhancement task. It is easily adaptable for mobile devices since it has only 47 thousand parameters. It shows the best SSIM 0.919 on RANDOM250 (MIT Adobe FiveK) among small models and is thrice faster than previous models.

Keywords: universal filter, image enhancement, neural networks, computer vision

Procedia PDF Downloads 66

Authors: Hyunbin Choi, Jihyeon Noh, Changwon Lim

Abstract:

In this paper, we propose SiamMask++, an architecture that performs layer-wise aggregation and depth-wise cross-correlation and introduce multi-RPN module and multi-MASK module to improve EAO (Expected Average Overlap), a representative performance evaluation metric for Visual Object Tracking (VOT) challenge. The proposed architecture, SiamMask++, has two versions, namely, bi_SiamMask++, which satisfies the real time (56fps) on systems equipped with GPUs (Titan XP), and rf_SiamMask++, which combines mask refinement modules for EAO improvements. Tests are performed on VOT2016, VOT2018 and VOT2019, the representative datasets of Visual Object Tracking tasks labeled as rotated bounding boxes. SiamMask++ perform better than SiamMask on all the three datasets tested. SiamMask++ is achieved performance of 62.6% accuracy, 26.2% robustness and 39.8% EAO, especially on the VOT2018 dataset. Compared to SiamMask, this is an improvement of 4.18%, 37.17%, 23.99%, respectively. In addition, we do an experimental in-depth analysis of how much the introduction of features and multi modules extracted from the backbone affects the performance of our model in the VOT task.

Keywords: visual object tracking, video, deep learning, layer wise aggregation, Siamese network

Procedia PDF Downloads 113

Authors: Sebastian Kölbl, Thomas Reitmaier, Mathias Hartmann

Abstract:

Trial and error approaches to adapt wearable support structures to human physiology are time consuming and elaborate. However, during preliminary design, the focus lies on understanding the interaction between exoskeleton and the human body in terms of forces and moments, namely body mechanics. For the study at hand, a multi-body simulation approach has been enhanced to evaluate actual forces and moments in a human dummy model with and without a digital mock-up of an active exoskeleton. Therefore, different motion data have been gathered and processed to perform a musculosceletal analysis. The motion data are ground reaction forces, electromyography data (EMG) and human motion data recorded with a marker-based motion capture system. Based on the experimental data, the response of the human dummy model has been calibrated. Subsequently, the scalable human dummy model, in conjunction with the motion data, is connected with the exoskeleton structure. The results of the human-machine interaction (HMI) simulation platform are in particular resulting contact forces and human joint forces to compare with admissible values with regard to the human physiology. Furthermore, it provides feedback for the sizing of the exoskeleton structure in terms of resulting interface forces (stress justification) and the effect of its compliance. A stepwise approach for the setup and validation of the modeling strategy is presented and the potential for a more time and cost-effective development of wearable support structures is outlined.

Keywords: assistive devices, ergonomic design, inverse dynamics, inverse kinematics, multibody simulation

Procedia PDF Downloads 125

Authors: Nemes Alexandra

Abstract:

The aim of the research is to develop an optimum microstructure of steel coatings on aluminum surfaces for application on the crankcase cylinder bores. For the proper design of the microstructure of the coat, it is important to control the plasma gun unit properly. The maximum operating time was determined while the plasma gun could optimally work before its destruction. Objectives: The aim of the research is to determine the optimal operating time of the plasma gun between renovations (the renovation shall involve the replacement of the test components of the plasma gun: electrode, nozzle plate, powder injector. Methodology: Plasma jet and particle flux analysis with PFI (PFI is a diagnostic tool for all kinds of thermal spraying processes), CT reconstruction and analysis on the new and the used plasma guns, failure analysis of electrodes, nozzle plates, and powder injectors, microscopic examination of the microstructure of the coating. Contributions: As the result of the failure analysis detailed above, the use of the plasma gun was maximized at 100 operating hours in order to get optimal microstructure for the coat.

Keywords: APS, air plasma spray, failure analysis, electrode, nozzle plate, powder injector

Procedia PDF Downloads 74

Authors: Julio Vega

Abstract:

Expression recognition is a field of Artificial Intelligence whose main objectives are to recognize basic forms of affective expression that appear on people’s faces and contributing to behavioral studies. In this work, a ROS node has been developed that, based on Deep Learning techniques, is capable of detecting the facial expressions of the people that appear in the image. These algorithms were optimized so that they can be executed in real time on an embedded platform. The experiments were carried out in a PC with a USB camera and in a Raspberry Pi 4 with a PiCamera. The final results shows a plausible system, which is capable to work in real time even in an embedded platform.

Keywords: python, low-cost, raspberry pi, emotion detection, human-robot interaction, ROS node

Procedia PDF Downloads 93

Authors: S. Zhuiykov, M. Karbalaei Akbari

Abstract:

Ultra-low-level incorporation of trace impurities and dopants into two-dimensional (2D) semiconductors is a challenging step towards the development of functional electronic instruments based on 2D materials. Herein, the incorporation of sulphur atoms into 2D Ga2O3 surface oxide film of eutectic gallium-indium alloy (EGaIn) is achieved through plasma-enhanced metal-catalyst dissociation of H2S gas on EGaIn substrate. This process led to the growth of GaS crystalline nanodomains inside amorphous 2D Ga2O3 sublayer films. Consequently, 2D lateral heterophase was developed between the amorphous Ga2O3 and crystalline GaS nanodomains. The materials characterization revealed the alteration of photoluminescence (PL) characteristics and change of valence band maximum (VBM) of functionalized 2D films. The comprehensive studies by conductive atomic force microscopy (c-AFM) showed considerable enhancement of conductivity of 2D Ga2O3/GaS materials (300 times improvement) compared with that of 2D Ga2O3 film. This technique has a great potential for the fabrication of 2D metal oxide devices with tuneable electronic characteristics similar to nano junction memristors and transistors.

Keywords: 2D semiconductors, Ga₂O₃, GaS, plasma-induced functionalization

Procedia PDF Downloads 54

Authors: Roghieh Eskandari, Ali Motamedi

Abstract:

There is a significant demand for renovation as-built assets are aging. To plan for a desirable and comfortable indoor environment, stakeholders use simulation technics to assess potential renovation scenarios with the innovative designs. Diminished Reality (DR), which is a technique of visually removing unwanted objects from the real-world scene in real-time, can contribute to the renovation design visualization for stakeholders by removing existing structures and assets from the scene. Using DR, the objects to be demolished or changed will be visually removed from the scene for a better understanding of the intended design scenarios for stakeholders. This research proposes an integrated system for renovation plan visualization using Building Information Modelling (BIM) data and mixed reality (MR) technologies. It presents a BIM-based DR method that utilizes a textured BIM model of the environment to accurately register the virtual model of the occluded background to the physical world in real-time. This system can facilitate the simulation of the renovation plan by visually diminishing building elements in an indoor environment.

Keywords: diminished reality, building information modelling, mixed reality, stock renovation

Procedia PDF Downloads 80

Authors: Omar Elharrouss, Youssef Hmamouche, Assia Kamal Idrissi, Btissam El Khamlichi, Amal El Fallah-Seghrouchni

Abstract:

Edge detection is represented as one of the most challenging tasks in computer vision, due to the complexity of detecting the edges or boundaries in real-world images that contains objects of different types and scales like trees, building as well as various backgrounds. Edge detection is represented also as a key task for many computer vision applications. Using a set of backbones as well as attention modules, deep-learning-based methods improved the detection of edges compared with the traditional methods like Sobel and Canny. However, images of complex scenes still represent a challenge for these methods. Also, the detected edges using the existing approaches suffer from non-refined results while the image output contains many erroneous edges. To overcome this, n this paper, by using the mechanism of residual learning, a refined edge detection network is proposed (RED-Net). By maintaining the high resolution of edges during the training process, and conserving the resolution of the edge image during the network stage, we make the pooling outputs at each stage connected with the output of the previous layer. Also, after each layer, we use an affined batch normalization layer as an erosion operation for the homogeneous region in the image. The proposed methods are evaluated using the most challenging datasets including BSDS500, NYUD, and Multicue. The obtained results outperform the designed edge detection networks in terms of performance metrics and quality of output images.

Keywords: edge detection, convolutional neural networks, deep learning, scale-representation, backbone

Procedia PDF Downloads 65

Authors: Omer Cahana, Ofer Levi, Maya Herman

Abstract:

Magnetic Resonance Imaging (MRI) is a lengthy medical scan that stems from a long acquisition time. Its length is mainly due to the traditional sampling theorem, which defines a lower boundary for sampling. However, it is still possible to accelerate the scan by using a different approach such as Compress Sensing (CS) or Parallel Imaging (PI). These two complementary methods can be combined to achieve a faster scan with high-fidelity imaging. To achieve that, two conditions must be satisfied: i) the signal must be sparse under a known transform domain, and ii) the sampling method must be incoherent. In addition, a nonlinear reconstruction algorithm must be applied to recover the signal. While the rapid advances in Deep Learning (DL) have had tremendous successes in various computer vision tasks, the field of MRI reconstruction is still in its early stages. In this paper, we present an end-to-end method for MRI reconstruction from k-space to image. Our method contains two parts. The first is sensitivity map estimation (SME), which is a small yet effective network that can easily be extended to a variable number of coils. The second is reconstruction, which is a top-down architecture with lateral connections developed for building high-level refinement at all scales. Our method holds the state-of-art fastMRI benchmark, which is the largest, most diverse benchmark for MRI reconstruction.

Keywords: magnetic resonance imaging, image reconstruction, pyramid network, deep learning

Procedia PDF Downloads 53

Authors: Lucy Jackson, Celyn Walters, Steve Eckersley, Mini Rai, Simon Hadfield

Abstract:

Space robots will play an integral part in exploring the universe and beyond. A correctly designed space robot will facilitate OOA, satellite servicing and ADR. However, problems arise when trying to design such a system as it is a highly complex multidimensional problem into which there is little research. Current design techniques are slow and specific to terrestrial manipulators. This paper presents a solution to the slow speed of robotic hardware design, and generalizes the technique to free-flying space robots. It presents Ta-DAH Design, an automated design approach that utilises a multi-objective cost function in an iterative and automated pipeline. The design approach leverages prior knowledge and facilitates the faster output of optimal designs. The result is a system that can optimise the size of the base spacecraft, manipulator and some key subsystems for any given task. Presented in this work is the methodology behind Ta-DAH Design and a number optimal space robot designs.

Keywords: space robots, automated design, on-orbit operations, hardware design

Procedia PDF Downloads 37

Authors: Krishna Mohan Bathula, Fatou Bintou Loucoubar, FNU Kaleemunnisa, Christelle Scharff, Mark Anthony De Castro

Abstract:

Voice recognition algorithms such as automatic speech recognition and text-to-speech systems with African languages can play an important role in bridging the digital divide of Artificial Intelligence in Africa, contributing to the establishment of a fully inclusive information society. This paper proposes a Deep Learning model that can classify the user responses as inputs for an interactive voice response system. A dataset with Wolof language words ‘yes’ and ‘no’ is collected as audio recordings. A two stage Data Augmentation approach is adopted for enhancing the dataset size required by the deep neural network. Data preprocessing and feature engineering with Mel-Frequency Cepstral Coefficients are implemented. Convolutional Neural Networks (CNNs) have proven to be very powerful in image classification and are promising for audio processing when sounds are transformed into spectra. For performing voice response classification, the recordings are transformed into sound frequency feature spectra and then applied image classification methodology using a deep CNN model. The inference model of this trained and reusable Wolof voice response recognition system can be integrated with many applications associated with both web and mobile platforms.

Keywords: automatic speech recognition, interactive voice response, voice response recognition, wolof word classification

Procedia PDF Downloads 80