Search results for: object detection with multimodal inputs
5103 Analysing Techniques for Fusing Multimodal Data in Predictive Scenarios Using Convolutional Neural Networks
Authors: Philipp Ruf, Massiwa Chabbi, Christoph Reich, Djaffar Ould-Abdeslam
Abstract:
In recent years, convolutional neural networks (CNN) have demonstrated high performance in image analysis, but oftentimes, there is only structured data available regarding a specific problem. By interpreting structured data as images, CNNs can effectively learn and extract valuable insights from tabular data, leading to improved predictive accuracy and uncovering hidden patterns that may not be apparent in traditional structured data analysis. In applying a single neural network for analyzing multimodal data, e.g., both structured and unstructured information, significant advantages in terms of time complexity and energy efficiency can be achieved. Converting structured data into images and merging them with existing visual material offers a promising solution for applying CNN in multimodal datasets, as they often occur in a medical context. By employing suitable preprocessing techniques, structured data is transformed into image representations, where the respective features are expressed as different formations of colors and shapes. In an additional step, these representations are fused with existing images to incorporate both types of information. This final image is finally analyzed using a CNN.Keywords: CNN, image processing, tabular data, mixed dataset, data transformation, multimodal fusion
Procedia PDF Downloads 1105102 Advancing in Cricket Analytics: Novel Approaches for Pitch and Ball Detection Employing OpenCV and YOLOV8
Authors: Pratham Madnur, Prathamkumar Shetty, Sneha Varur, Gouri Parashetti
Abstract:
In order to overcome conventional obstacles, this research paper investigates novel approaches for cricket pitch and ball detection that make use of cutting-edge technologies. The research integrates OpenCV for pitch inspection and modifies the YOLOv8 model for cricket ball detection in order to overcome the shortcomings of manual pitch assessment and traditional ball detection techniques. To ensure flexibility in a range of pitch environments, the pitch detection method leverages OpenCV’s color space transformation, contour extraction, and accurate color range defining features. Regarding ball detection, the YOLOv8 model emphasizes the preservation of minor object details to improve accuracy and is specifically trained to the unique properties of cricket balls. The methods are more reliable because of the careful preparation of the datasets, which include novel ball and pitch information. These cutting-edge methods not only improve cricket analytics but also set the stage for flexible methods in more general sports technology applications.Keywords: OpenCV, YOLOv8, cricket, custom dataset, computer vision, sports
Procedia PDF Downloads 615101 Stereo Motion Tracking
Authors: Yudhajit Datta, Hamsi Iyer, Jonathan Bandi, Ankit Sethia
Abstract:
Motion Tracking and Stereo Vision are complicated, albeit well-understood problems in computer vision. Existing softwares that combine the two approaches to perform stereo motion tracking typically employ complicated and computationally expensive procedures. The purpose of this study is to create a simple and effective solution capable of combining the two approaches. The study aims to explore a strategy to combine the two techniques of two-dimensional motion tracking using Kalman Filter; and depth detection of object using Stereo Vision. In conventional approaches objects in the scene of interest are observed using a single camera. However for Stereo Motion Tracking; the scene of interest is observed using video feeds from two calibrated cameras. Using two simultaneous measurements from the two cameras a calculation for the depth of the object from the plane containing the cameras is made. The approach attempts to capture the entire three-dimensional spatial information of each object at the scene and represent it through a software estimator object. In discrete intervals, the estimator tracks object motion in the plane parallel to plane containing cameras and updates the perpendicular distance value of the object from the plane containing the cameras as depth. The ability to efficiently track the motion of objects in three-dimensional space using a simplified approach could prove to be an indispensable tool in a variety of surveillance scenarios. The approach may find application from high security surveillance scenes such as premises of bank vaults, prisons or other detention facilities; to low cost applications in supermarkets and car parking lots.Keywords: kalman filter, stereo vision, motion tracking, matlab, object tracking, camera calibration, computer vision system toolbox
Procedia PDF Downloads 3235100 Stabilization Technique for Multi-Inputs Voltage Sense Amplifiers in Node Sharing Converters
Authors: Sanghoon Park, Ki-Jin Kim, Kwang-Ho Ahn
Abstract:
This paper discusses the undesirable charge transfer through the parasitic capacitances of the input transistors in a multi-inputs voltage sense amplifier. Its intrinsic rail-to-rail voltage transitions at the output nodes inevitably disturb the input sides through the capacitive coupling between the outputs and inputs. Then, it can possible degrade the stabilities of the reference voltage levels. Moreover, it becomes more serious in multi-channel systems by altering them for other channels, and so degrades the linearity of the overall systems. In order to alleviate the internal node voltage transition, the internal node stabilization techniques are proposed. It achieves 45% and 40% improvements for node stabilization and input referred disturbance, respectively.Keywords: voltage sense amplifier, multi-inputs, voltage transition, node stabilization, biasing circuits
Procedia PDF Downloads 5585099 Effective Stacking of Deep Neural Models for Automated Object Recognition in Retail Stores
Authors: Ankit Sinha, Soham Banerjee, Pratik Chattopadhyay
Abstract:
Automated product recognition in retail stores is an important real-world application in the domain of Computer Vision and Pattern Recognition. In this paper, we consider the problem of automatically identifying the classes of the products placed on racks in retail stores from an image of the rack and information about the query/product images. We improve upon the existing approaches in terms of effectiveness and memory requirement by developing a two-stage object detection and recognition pipeline comprising of a Faster-RCNN-based object localizer that detects the object regions in the rack image and a ResNet-18-based image encoder that classifies the detected regions into the appropriate classes. Each of the models is fine-tuned using appropriate data sets for better prediction and data augmentation is performed on each query image to prepare an extensive gallery set for fine-tuning the ResNet-18-based product recognition model. This encoder is trained using a triplet loss function following the strategy of online-hard-negative-mining for improved prediction. The proposed models are lightweight and can be connected in an end-to-end manner during deployment to automatically identify each product object placed in a rack image. Extensive experiments using Grozi-32k and GP-180 data sets verify the effectiveness of the proposed model.Keywords: retail stores, faster-RCNN, object localization, ResNet-18, triplet loss, data augmentation, product recognition
Procedia PDF Downloads 1455098 Low Cost Real Time Robust Identification of Impulsive Signals
Authors: R. Biondi, G. Dys, G. Ferone, T. Renard, M. Zysman
Abstract:
This paper describes an automated implementable system for impulsive signals detection and recognition. The system uses a Digital Signal Processing device for the detection and identification process. Here the system analyses the signals in real time in order to produce a particular response if needed. The system analyses the signals in real time in order to produce a specific output if needed. Detection is achieved through normalizing the inputs and comparing the read signals to a dynamic threshold and thus avoiding detections linked to loud or fluctuating environing noise. Identification is done through neuronal network algorithms. As a setup our system can receive signals to “learn” certain patterns. Through “learning” the system can recognize signals faster, inducing flexibility to new patterns similar to those known. Sound is captured through a simple jack input, and could be changed for an enhanced recording surface such as a wide-area recorder. Furthermore a communication module can be added to the apparatus to send alerts to another interface if needed.Keywords: sound detection, impulsive signal, background noise, neural network
Procedia PDF Downloads 3125097 Dual Biometrics Fusion Based Recognition System
Authors: Prakash, Vikash Kumar, Vinay Bansal, L. N. Das
Abstract:
Dual biometrics is a subpart of multimodal biometrics, which refers to the use of a variety of modalities to identify and authenticate persons rather than just one. We limit the risks of mistakes by mixing several modals, and hackers have a tiny possibility of collecting information. Our goal is to collect the precise characteristics of iris and palmprint, produce a fusion of both methodologies, and ensure that authentication is only successful when the biometrics match a particular user. After combining different modalities, we created an effective strategy with a mean DI and EER of 2.41 and 5.21, respectively. A biometric system has been proposed.Keywords: multimodal, fusion, palmprint, Iris, EER, DI
Procedia PDF Downloads 1425096 Object-Oriented Program Comprehension by Identification of Software Components and Their Connexions
Authors: Abdelhak-Djamel Seriai, Selim Kebir, Allaoua Chaoui
Abstract:
During the last decades, object oriented program- ming has been massively used to build large-scale systems. However, evolution and maintenance of such systems become a laborious task because of the lack of object oriented programming to offer a precise view of the functional building blocks of the system. This lack is caused by the fine granularity of classes and objects. In this paper, we use a post object-oriented technology namely software components, to propose an approach based on the identification of the functional building blocks of an object oriented system by analyzing its source code. These functional blocks are specified as software components and the result is a multi-layer component based software architecture.Keywords: software comprehension, software component, object oriented, software architecture, reverse engineering
Procedia PDF Downloads 4075095 An Image Processing Scheme for Skin Fungal Disease Identification
Authors: A. A. M. A. S. S. Perera, L. A. Ranasinghe, T. K. H. Nimeshika, D. M. Dhanushka Dissanayake, Namalie Walgampaya
Abstract:
Nowadays, skin fungal diseases are mostly found in people of tropical countries like Sri Lanka. A skin fungal disease is a particular kind of illness caused by fungus. These diseases have various dangerous effects on the skin and keep on spreading over time. It becomes important to identify these diseases at their initial stage to control it from spreading. This paper presents an automated skin fungal disease identification system implemented to speed up the diagnosis process by identifying skin fungal infections in digital images. An image of the diseased skin lesion is acquired and a comprehensive computer vision and image processing scheme is used to process the image for the disease identification. This includes colour analysis using RGB and HSV colour models, texture classification using Grey Level Run Length Matrix, Grey Level Co-Occurrence Matrix and Local Binary Pattern, Object detection, Shape Identification and many more. This paper presents the approach and its outcome for identification of four most common skin fungal infections, namely, Tinea Corporis, Sporotrichosis, Malassezia and Onychomycosis. The main intention of this research is to provide an automated skin fungal disease identification system that increase the diagnostic quality, shorten the time-to-diagnosis and improve the efficiency of detection and successful treatment for skin fungal diseases.Keywords: Circularity Index, Grey Level Run Length Matrix, Grey Level Co-Occurrence Matrix, Local Binary Pattern, Object detection, Ring Detection, Shape Identification
Procedia PDF Downloads 2265094 Self-Organizing Maps for Credit Card Fraud Detection
Authors: ChunYi Peng, Wei Hsuan CHeng, Shyh Kuang Ueng
Abstract:
This study focuses on the application of self-organizing maps (SOM) technology in analyzing credit card transaction data, aiming to enhance the accuracy and efficiency of fraud detection. Som, as an artificial neural network, is particularly suited for pattern recognition and data classification, making it highly effective for the complex and variable nature of credit card transaction data. By analyzing transaction characteristics with SOM, the research identifies abnormal transaction patterns that could indicate potentially fraudulent activities. Moreover, this study has developed a specialized visualization tool to intuitively present the relationships between SOM analysis outcomes and transaction data, aiding financial institution personnel in quickly identifying and responding to potential fraud, thereby reducing financial losses. Additionally, the research explores the integration of SOM technology with composite intelligent system technologies (including finite state machines, fuzzy logic, and decision trees) to further improve fraud detection accuracy. This multimodal approach provides a comprehensive perspective for identifying and understanding various types of fraud within credit card transactions. In summary, by integrating SOM technology with visualization tools and composite intelligent system technologies, this research offers a more effective method of fraud detection for the financial industry, not only enhancing detection accuracy but also deepening the overall understanding of fraudulent activities.Keywords: self-organizing map technology, fraud detection, information visualization, data analysis, composite intelligent system technologies, decision support technologies
Procedia PDF Downloads 455093 Image Instance Segmentation Using Modified Mask R-CNN
Authors: Avatharam Ganivada, Krishna Shah
Abstract:
The Mask R-CNN is recently introduced by the team of Facebook AI Research (FAIR), which is mainly concerned with instance segmentation in images. Here, the Mask R-CNN is based on ResNet and feature pyramid network (FPN), where a single dropout method is employed. This paper provides a modified Mask R-CNN by adding multiple dropout methods into the Mask R-CNN. The proposed model has also utilized the concepts of Resnet and FPN to extract stage-wise network feature maps, wherein a top-down network path having lateral connections is used to obtain semantically strong features. The proposed model produces three outputs for each object in the image: class label, bounding box coordinates, and object mask. The performance of the proposed network is evaluated in the segmentation of every instance in images using COCO and cityscape datasets. The proposed model achieves better performance than the state-of-the-networks for the datasets.Keywords: instance segmentation, object detection, convolutional neural networks, deep learning, computer vision
Procedia PDF Downloads 675092 UAV Based Visual Object Tracking
Authors: Vaibhav Dalmia, Manoj Phirke, Renith G
Abstract:
With the wide adoption of UAVs (unmanned aerial vehicles) in various industries by the government as well as private corporations for solving computer vision tasks it’s necessary that their potential is analyzed completely. Recent advances in Deep Learning have also left us with a plethora of algorithms to solve different computer vision tasks. This study provides a comprehensive survey on solving the Visual Object Tracking problem and explains the tradeoffs involved in building a real-time yet reasonably accurate object tracking system for UAVs by looking at existing methods and evaluating them on the aerial datasets. Finally, the best trackers suitable for UAV-based applications are provided.Keywords: deep learning, drones, single object tracking, visual object tracking, UAVs
Procedia PDF Downloads 1485091 Rehabilitation of the Blind Using Sono-Visualization Tool
Authors: Ashwani Kumar
Abstract:
In human beings, eyes play a vital role. A very less research has been done for rehabilitation of blindness for the blind people. This paper discusses the work that helps blind people for recognizing the basic shapes of the objects like circle, square, triangle, horizontal lines, vertical lines, diagonal lines and the wave forms like sinusoidal, square, triangular etc. This is largely achieved by using a digital camera, which is used to capture the visual information present in front of the blind person and a software program, which achieves the image processing operations, and finally the processed image is converted into sound. After the sound generation process, the generated sound is fed to the blind person through headphones for visualizing the imaginary image of the object. For visualizing the imaginary image of the object, it needs to train the blind person. Various training process methods had been applied for recognizing the object.Keywords: image processing, pixel, pitch, loudness, sound generation, edge detection, brightness
Procedia PDF Downloads 3785090 Self-Organizing Maps for Credit Card Fraud Detection and Visualization
Authors: Peng Chun-Yi, Chen Wei-Hsuan, Ueng Shyh-Kuang
Abstract:
This study focuses on the application of self-organizing maps (SOM) technology in analyzing credit card transaction data, aiming to enhance the accuracy and efficiency of fraud detection. Som, as an artificial neural network, is particularly suited for pattern recognition and data classification, making it highly effective for the complex and variable nature of credit card transaction data. By analyzing transaction characteristics with SOM, the research identifies abnormal transaction patterns that could indicate potentially fraudulent activities. Moreover, this study has developed a specialized visualization tool to intuitively present the relationships between SOM analysis outcomes and transaction data, aiding financial institution personnel in quickly identifying and responding to potential fraud, thereby reducing financial losses. Additionally, the research explores the integration of SOM technology with composite intelligent system technologies (including finite state machines, fuzzy logic, and decision trees) to further improve fraud detection accuracy. This multimodal approach provides a comprehensive perspective for identifying and understanding various types of fraud within credit card transactions. In summary, by integrating SOM technology with visualization tools and composite intelligent system technologies, this research offers a more effective method of fraud detection for the financial industry, not only enhancing detection accuracy but also deepening the overall understanding of fraudulent activities.Keywords: self-organizing map technology, fraud detection, information visualization, data analysis, composite intelligent system technologies, decision support technologies
Procedia PDF Downloads 535089 Object-Oriented Modeling Simulation and Control of Activated Sludge Process
Authors: J. Fernandez de Canete, P. Del Saz Orozco, I. Garcia-Moral, A. Akhrymenka
Abstract:
Object-oriented modeling is spreading in current simulation of wastewater treatments plants through the use of the individual components of the process and its relations to define the underlying dynamic equations. In this paper, we describe the use of the free-software OpenModelica simulation environment for the object-oriented modeling of an activated sludge process under feedback control. The performance of the controlled system was analyzed both under normal conditions and in the presence of disturbances. The object-oriented described approach represents a valuable tool in teaching provides a practical insight in wastewater process control field.Keywords: object-oriented programming, activated sludge process, OpenModelica, feedback control
Procedia PDF Downloads 3795088 Subpixel Corner Detection for Monocular Camera Linear Model Research
Authors: Guorong Sui, Xingwei Jia, Fei Tong, Xiumin Gao
Abstract:
Camera calibration is a fundamental issue of high precision noncontact measurement. And it is necessary to analyze and study the reliability and application range of its linear model which is often used in the camera calibration. According to the imaging features of monocular cameras, a camera model which is based on the image pixel coordinates and three dimensional space coordinates is built. Using our own customized template, the image pixel coordinate is obtained by the subpixel corner detection method. Without considering the aberration of the optical system, the feature extraction and linearity analysis of the line segment in the template are performed. Moreover, the experiment is repeated 11 times by constantly varying the measuring distance. At last, the linearity of the camera is achieved by fitting 11 groups of data. The camera model measurement results show that the relative error does not exceed 1%, and the repeated measurement error is not more than 0.1 mm magnitude. Meanwhile, it is found that the model has some measurement differences in the different region and object distance. The experiment results show this linear model is simple and practical, and have good linearity within a certain object distance. These experiment results provide a powerful basis for establishment of the linear model of camera. These works will have potential value to the actual engineering measurement.Keywords: camera linear model, geometric imaging relationship, image pixel coordinates, three dimensional space coordinates, sub-pixel corner detection
Procedia PDF Downloads 2715087 Deep Learning Based Road Crack Detection on an Embedded Platform
Authors: Nurhak Altın, Ayhan Kucukmanisa, Oguzhan Urhan
Abstract:
It is important that highways are in good condition for traffic safety. Road crashes (road cracks, erosion of lane markings, etc.) can cause accidents by affecting driving. Image processing based methods for detecting road cracks are available in the literature. In this paper, a deep learning based road crack detection approach is proposed. YOLO (You Look Only Once) is adopted as core component of the road crack detection approach presented. The YOLO network structure, which is developed for object detection, is trained with road crack images as a new class that is not previously used in YOLO. The performance of the proposed method is compared using different training methods: using randomly generated weights and training their own pre-trained weights (transfer learning). A similar training approach is applied to the simplified version of the YOLO network model (tiny yolo) and the results of the performance are examined. The developed system is able to process 8 fps on NVIDIA Jetson TX1 development kit.Keywords: deep learning, embedded platform, real-time processing, road crack detection
Procedia PDF Downloads 3325086 Separating Permanent and Induced Magnetic Signature: A Simple Approach
Authors: O. J. G. Somsen, G. P. M. Wagemakers
Abstract:
Magnetic signature detection provides sensitive detection of metal objects, especially in the natural environment. Our group is developing a tabletop setup for magnetic signatures of various small and model objects. A particular issue is the separation of permanent and induced magnetization. While the latter depends only on the composition and shape of the object, the former also depends on the magnetization history. With common deperming techniques, a significant permanent signature may still remain, which confuses measurements of the induced component. We investigate a basic technique of separating the two. Measurements were done by moving the object along an aluminum rail while the three field components are recorded by a detector attached near the center. This is done first with the rail parallel to the Earth magnetic field and then with anti-parallel orientation. The reversal changes the sign of the induced- but not the permanent magnetization so that the two can be separated. Our preliminary results on a small iron block show excellent reproducibility. A considerable permanent magnetization was indeed present, resulting in a complex asymmetric signature. After separation, a much more symmetric induced signature was obtained that can be studied in detail and compared with theoretical calculations.Keywords: magnetic signature, data analysis, magnetization, deperming techniques
Procedia PDF Downloads 4475085 Improving the Analytical Power of Dynamic DEA Models, by the Consideration of the Shape of the Distribution of Inputs/Outputs Data: A Linear Piecewise Decomposition Approach
Authors: Elias K. Maragos, Petros E. Maravelakis
Abstract:
In Dynamic Data Envelopment Analysis (DDEA), which is a subfield of Data Envelopment Analysis (DEA), the productivity of Decision Making Units (DMUs) is considered in relation to time. In this case, as it is accepted by the most of the researchers, there are outputs, which are produced by a DMU to be used as inputs in a future time. Those outputs are known as intermediates. The common models, in DDEA, do not take into account the shape of the distribution of those inputs, outputs or intermediates data, assuming that the distribution of the virtual value of them does not deviate from linearity. This weakness causes the limitation of the accuracy of the analytical power of the traditional DDEA models. In this paper, the authors, using the concept of piecewise linear inputs and outputs, propose an extended DDEA model. The proposed model increases the flexibility of the traditional DDEA models and improves the measurement of the dynamic performance of DMUs.Keywords: Dynamic Data Envelopment Analysis, DDEA, piecewise linear inputs, piecewise linear outputs
Procedia PDF Downloads 1555084 A Multimodal Approach to Improve the Performance of Biometric System
Authors: Chander Kant, Arun Kumar
Abstract:
Biometric systems automatically recognize an individual based on his/her physiological and behavioral characteristics. There are also some traits like weight, age, height etc. that may not provide reliable user recognition because of there common and temporary nature. These traits are called soft bio metric traits. Although soft bio metric traits are lack of permanence to uniquely and reliably identify an individual, yet they provide some beneficial evidence about the user identity and may improve the system performance. Here in this paper, we have proposed an approach for integrating the soft bio metrics with fingerprint and face to improve the performance of personal authentication system. In our approach we have proposed a combined architecture of three different sensors to elevate the system performance. The approach includes, soft bio metrics, fingerprint and face traits. We have also proven the efficiency of proposed system regarding FAR (False Acceptance Ratio) and total response time, with the help of MUBI (Multimodal Bio metrics Integration) software.Keywords: FAR, minutiae point, multimodal bio metrics, primary bio metric, soft bio metric
Procedia PDF Downloads 3385083 Hedgerow Detection and Characterization Using Very High Spatial Resolution SAR DATA
Authors: Saeid Gharechelou, Stuart Green, Fiona Cawkwell
Abstract:
Hedgerow has an important role for a wide range of ecological habitats, landscape, agriculture management, carbon sequestration, wood production. Hedgerow detection accurately using satellite imagery is a challenging problem in remote sensing techniques, because in the special approach it is very similar to line object like a road, from a spectral viewpoint, a hedge is very similar to a forest. Remote sensors with very high spatial resolution (VHR) recently enable the automatic detection of hedges by the acquisition of images with enough spectral and spatial resolution. Indeed, recently VHR remote sensing data provided the opportunity to detect the hedgerow as line feature but still remain difficulties in monitoring the characterization in landscape scale. In this research is used the TerraSAR-x Spotlight and Staring mode with 3-5 m resolution in wet and dry season in the test site of Fermoy County, Ireland to detect the hedgerow by acquisition time of 2014-2015. Both dual polarization of Spotlight data in HH/VV is using for detection of hedgerow. The varied method of SAR image technique with try and error way by integration of classification algorithm like texture analysis, support vector machine, k-means and random forest are using to detect hedgerow and its characterization. We are applying the Shannon entropy (ShE) and backscattering analysis in single and double bounce in polarimetric analysis for processing the object-oriented classification and finally extracting the hedgerow network. The result still is in progress and need to apply the other method as well to find the best method in study area. Finally, this research is under way to ahead to get the best result and here just present the preliminary work that polarimetric image of TSX potentially can detect the hedgerow.Keywords: TerraSAR-X, hedgerow detection, high resolution SAR image, dual polarization, polarimetric analysis
Procedia PDF Downloads 2275082 Training of Future Computer Science Teachers Based on Machine Learning Methods
Authors: Meruert Serik, Nassipzhan Duisegaliyeva, Danara Tleumagambetova
Abstract:
The article highlights and describes the characteristic features of real-time face detection in images and videos using machine learning algorithms. Students of educational programs reviewed the research work "6B01511-Computer Science", "7M01511-Computer Science", "7M01525- STEM Education," and "8D01511-Computer Science" of Eurasian National University named after L.N. Gumilyov. As a result, the advantages and disadvantages of Haar Cascade (Haar Cascade OpenCV), HoG SVM (Histogram of Oriented Gradients, Support Vector Machine), and MMOD CNN Dlib (Max-Margin Object Detection, convolutional neural network) detectors used for face detection were determined. Dlib is a general-purpose cross-platform software library written in the programming language C++. It includes detectors used for determining face detection. The Cascade OpenCV algorithm is efficient for fast face detection. The considered work forms the basis for the development of machine learning methods by future computer science teachers.Keywords: algorithm, artificial intelligence, education, machine learning
Procedia PDF Downloads 685081 Multiperson Drone Control with Seamless Pilot Switching Using Onboard Camera and Openpose Real-Time Keypoint Detection
Authors: Evan Lowhorn, Rocio Alba-Flores
Abstract:
Traditional classification Convolutional Neural Networks (CNN) attempt to classify an image in its entirety. This becomes problematic when trying to perform classification with a drone’s camera in real-time due to unpredictable backgrounds. Object detectors with bounding boxes can be used to isolate individuals and other items, but the original backgrounds remain within these boxes. These basic detectors have been regularly used to determine what type of object an item is, such as “person” or “dog.” Recent advancement in computer vision, particularly with human imaging, is keypoint detection. Human keypoint detection goes beyond bounding boxes to fully isolate humans and plot points, or Regions of Interest (ROI), on their bodies within an image. ROIs can include shoulders, elbows, knees, heads, etc. These points can then be related to each other and used in deep learning methods such as pose estimation. For drone control based on human motions, poses, or signals using the onboard camera, it is important to have a simple method for pilot identification among multiple individuals while also giving the pilot fine control options for the drone. To achieve this, the OpenPose keypoint detection network was used with body and hand keypoint detection enabled. OpenPose supports the ability to combine multiple keypoint detection methods in real-time with a single network. Body keypoint detection allows simple poses to act as the pilot identifier. The hand keypoint detection with ROIs for each finger can then offer a greater variety of signal options for the pilot once identified. For this work, the individual must raise their non-control arm to be identified as the operator and send commands with the hand on their other arm. The drone ignores all other individuals in the onboard camera feed until the current operator lowers their non-control arm. When another individual wish to operate the drone, they simply raise their arm once the current operator relinquishes control, and then they can begin controlling the drone with their other hand. This is all performed mid-flight with no landing or script editing required. When using a desktop with a discrete NVIDIA GPU, the drone’s 2.4 GHz Wi-Fi connection combined with OpenPose restrictions to only body and hand allows this control method to perform as intended while maintaining the responsiveness required for practical use.Keywords: computer vision, drone control, keypoint detection, openpose
Procedia PDF Downloads 1785080 Filmic and Verbal Metafphors
Authors: Manana Rusieshvili, Rusudan Dolidze
Abstract:
This paper aims at 1) investigating the ways in which a traditional, monomodal written verbal metaphor can be transposed as a monomodal non-verbal (visual) or multimodal (aural and -visual) filmic metaphor ; 2) exploring similarities and differences in the process of encoding and decoding of monomodal and multimodal metaphors. The empiric data, on which the research is based, embrace three sources: the novel by Harry Gray ‘The Hoods’, the script of the film ‘Once Upon a Time in America’ (English version by David Mills) and the resultant film by Sergio Leone. In order to achieve the above mentioned goals, the research focuses on the following issues: 1) identification of verbal and non-verbal monomodal and multimodal metaphors in the above-mentioned sources and 2) investigation of the ways and modes the specific written monomodal metaphors appearing in the novel and the script are enacted in the film and become visual, aural or visual-aural filmic metaphors ; 3) study of the factors which play an important role in contributing to the encoding and decoding of the filmic metaphor. The collection and analysis of the data were carried out in two stages: firstly, the relevant data, i.e. the monomodal metaphors from the novel, the script and the film were identified and collected. In the second, final stage the metaphors taken from all of the three sources were analysed, compared and two types of phenomena were selected for discussion: (1) the monomodal written metaphors found in the novel and/or in the script which become monomodal visual/aural metaphors in the film; (2) the monomodal written metaphors found in the novel and/or in the script which become multimodal, filmic (visual-aural) metaphors in the film.Keywords: encoding, decoding, filmic metaphor, multimodality
Procedia PDF Downloads 5205079 On the Study of the Electromagnetic Scattering by Large Obstacle Based on the Method of Auxiliary Sources
Authors: Hidouri Sami, Aguili Taoufik
Abstract:
We consider fast and accurate solutions of scattering problems by large perfectly conducting objects (PEC) formulated by an optimization of the Method of Auxiliary Sources (MAS). We present various techniques used to reduce the total computational cost of the scattering problem. The first technique is based on replacing the object by an array of finite number of small (PEC) object with the same shape. The second solution reduces the problem on considering only the half of the object.These two solutions are compared to results from the reference bibliography.Keywords: method of auxiliary sources, scattering, large object, RCS, computational resources
Procedia PDF Downloads 2315078 Investigation of Detectability of Orbital Objects/Debris in Geostationary Earth Orbit by Microwave Kinetic Inductance Detectors
Authors: Saeed Vahedikamal, Ian Hepburn
Abstract:
Microwave Kinetic Inductance Detectors (MKIDs) are considered as one of the most promising photon detectors of the future in many Astronomical applications such as exoplanet detections. The MKID advantages stem from their single photon sensitivity (ranging from UV to optical and near infrared), photon energy resolution and high temporal capability (~microseconds). There has been substantial progress in the development of these detectors and MKIDs with Megapixel arrays is now possible. The unique capability of recording an incident photon and its energy (or wavelength) while also registering its time of arrival to within a microsecond enables an array of MKIDs to produce a four-dimensional data block of x, y, z and t comprising x, y spatial, z axis per pixel spectral and t axis per pixel which is temporal. This offers the possibility that the spectrum and brightness variation for any detected piece of space debris as a function of time might offer a unique identifier or fingerprint. Such a fingerprint signal from any object identified in multiple detections by different observers has the potential to determine the orbital features of the object and be used for their tracking. Modelling performed so far shows that with a 20 cm telescope located at an Astronomical observatory (e.g. La Palma, Canary Islands) we could detect sub cm objects at GEO. By considering a Lambertian sphere with a 10 % reflectivity (albedo of the Moon) we anticipate the following for a GEO object: 10 cm object imaged in a 1 second image capture; 1.2 cm object for a 70 second image integration or 0.65 cm object for a 4 minute image integration. We present details of our modelling and the potential instrument for a dedicated GEO surveillance system.Keywords: space debris, orbital debris, detection system, observation, microwave kinetic inductance detectors, MKID
Procedia PDF Downloads 885077 A Framework for Automated Nuclear Waste Classification
Authors: Seonaid Hume, Gordon Dobie, Graeme West
Abstract:
Detecting and localizing radioactive sources is a necessity for safe and secure decommissioning of nuclear facilities. An important aspect for the management of the sort-and-segregation process is establishing the spatial distributions and quantities of the waste radionuclides, their type, corresponding activity, and ultimately classification for disposal. The data received from surveys directly informs decommissioning plans, on-site incident management strategies, the approach needed for a new cell, as well as protecting the workforce and the public. Manual classification of nuclear waste from a nuclear cell is time-consuming, expensive, and requires significant expertise to make the classification judgment call. Also, in-cell decommissioning is still in its relative infancy, and few techniques are well-developed. As with any repetitive and routine tasks, there is the opportunity to improve the task of classifying nuclear waste using autonomous systems. Hence, this paper proposes a new framework for the automatic classification of nuclear waste. This framework consists of five main stages; 3D spatial mapping and object detection, object classification, radiological mapping, source localisation based on gathered evidence and finally, waste classification. The first stage of the framework, 3D visual mapping, involves object detection from point cloud data. A review of related applications in other industries is provided, and recommendations for approaches for waste classification are made. Object detection focusses initially on cylindrical objects since pipework is significant in nuclear cells and indeed any industrial site. The approach can be extended to other commonly occurring primitives such as spheres and cubes. This is in preparation of stage two, characterizing the point cloud data and estimating the dimensions, material, degradation, and mass of the objects detected in order to feature match them to an inventory of possible items found in that nuclear cell. Many items in nuclear cells are one-offs, have limited or poor drawings available, or have been modified since installation, and have complex interiors, which often and inadvertently pose difficulties when accessing certain zones and identifying waste remotely. Hence, this may require expert input to feature match objects. The third stage, radiological mapping, is similar in order to facilitate the characterization of the nuclear cell in terms of radiation fields, including the type of radiation, activity, and location within the nuclear cell. The fourth stage of the framework takes the visual map for stage 1, the object characterization from stage 2, and radiation map from stage 3 and fuses them together, providing a more detailed scene of the nuclear cell by identifying the location of radioactive materials in three dimensions. The last stage involves combining the evidence from the fused data sets to reveal the classification of the waste in Bq/kg, thus enabling better decision making and monitoring for in-cell decommissioning. The presentation of the framework is supported by representative case study data drawn from an application in decommissioning from a UK nuclear facility. This framework utilises recent advancements of the detection and mapping capabilities of complex radiation fields in three dimensions to make the process of classifying nuclear waste faster, more reliable, cost-effective and safer.Keywords: nuclear decommissioning, radiation detection, object detection, waste classification
Procedia PDF Downloads 1945076 Audio-Visual Co-Data Processing Pipeline
Authors: Rita Chattopadhyay, Vivek Anand Thoutam
Abstract:
Speech is the most acceptable means of communication where we can quickly exchange our feelings and thoughts. Quite often, people can communicate orally but cannot interact or work with computers or devices. It’s easy and quick to give speech commands than typing commands to computers. In the same way, it’s easy listening to audio played from a device than extract output from computers or devices. Especially with Robotics being an emerging market with applications in warehouses, the hospitality industry, consumer electronics, assistive technology, etc., speech-based human-machine interaction is emerging as a lucrative feature for robot manufacturers. Considering this factor, the objective of this paper is to design the “Audio-Visual Co-Data Processing Pipeline.” This pipeline is an integrated version of Automatic speech recognition, a Natural language model for text understanding, object detection, and text-to-speech modules. There are many Deep Learning models for each type of the modules mentioned above, but OpenVINO Model Zoo models are used because the OpenVINO toolkit covers both computer vision and non-computer vision workloads across Intel hardware and maximizes performance, and accelerates application development. A speech command is given as input that has information about target objects to be detected and start and end times to extract the required interval from the video. Speech is converted to text using the Automatic speech recognition QuartzNet model. The summary is extracted from text using a natural language model Generative Pre-Trained Transformer-3 (GPT-3). Based on the summary, essential frames from the video are extracted, and the You Only Look Once (YOLO) object detection model detects You Only Look Once (YOLO) objects on these extracted frames. Frame numbers that have target objects (specified objects in the speech command) are saved as text. Finally, this text (frame numbers) is converted to speech using text to speech model and will be played from the device. This project is developed for 80 You Only Look Once (YOLO) labels, and the user can extract frames based on only one or two target labels. This pipeline can be extended for more than two target labels easily by making appropriate changes in the object detection module. This project is developed for four different speech command formats by including sample examples in the prompt used by Generative Pre-Trained Transformer-3 (GPT-3) model. Based on user preference, one can come up with a new speech command format by including some examples of the respective format in the prompt used by the Generative Pre-Trained Transformer-3 (GPT-3) model. This pipeline can be used in many projects like human-machine interface, human-robot interaction, and surveillance through speech commands. All object detection projects can be upgraded using this pipeline so that one can give speech commands and output is played from the device.Keywords: OpenVINO, automatic speech recognition, natural language processing, object detection, text to speech
Procedia PDF Downloads 745075 Efficient Layout-Aware Pretraining for Multimodal Form Understanding
Authors: Armineh Nourbakhsh, Sameena Shah, Carolyn Rose
Abstract:
Layout-aware language models have been used to create multimodal representations for documents that are in image form, achieving relatively high accuracy in document understanding tasks. However, the large number of parameters in the resulting models makes building and using them prohibitive without access to high-performing processing units with large memory capacity. We propose an alternative approach that can create efficient representations without the need for a neural visual backbone. This leads to an 80% reduction in the number of parameters compared to the smallest SOTA model, widely expanding applicability. In addition, our layout embeddings are pre-trained on spatial and visual cues alone and only fused with text embeddings in downstream tasks, which can facilitate applicability to low-resource of multi-lingual domains. Despite using 2.5% of training data, we show competitive performance on two form understanding tasks: semantic labeling and link prediction.Keywords: layout understanding, form understanding, multimodal document understanding, bias-augmented attention
Procedia PDF Downloads 1415074 A Multimodal Approach towards Intersemiotic Translations of 'The Great Gatsby'
Authors: Neda Razavi Kaleibar, Bahloul Salmani
Abstract:
The present study dealt with the multimodal analysis of two cinematic adaptations of The Great Gatsby as intersemiotic translation. The assessment in this study went beyond the faithfulness based on repetition, addition, deletion, and creation which limit the analysis from other aspects. In fact, this research aimed to pinpoint the role of multimodality in examining the intersemiotic translations of the novel into film by means of analyzing different applied modes. Through a qualitative type of research, the analysis was conducted based on the theory proposed by Burn as Kineikonic mode theory derived from the concept of multimodality. The results of the study revealed that due to the applied modes, each adaptation represents a sense and meaning different from the other one. Analyzing the results and discussions, it was concluded that not only the modes have an undeniable role in film adaptations, but rather multimodal analysis including different nonverbal modes can be a useful and functional choice for analyzing the intersemiotic translations.Keywords: cinematic adaptation, intersemiotic translation, kineikonic mode, multimodality
Procedia PDF Downloads 410