Search results for: feature detection and description
Commenced in January 2007
Frequency: Monthly
Edition: International
Paper Count: 5433

Search results for: feature detection and description

4593 A Machine Learning Pipeline for Real-Time Activity Detection on Low Computational Power Devices for Metaverse Applications

Authors: Amit Kumar, Amanpreet Chander, Ashish Sahani

Abstract:

This paper presents our recent work on real-time human activity detection based on the media pipe pipeline and machine learning algorithms. The proposed system can detect human activities, including running, jumping, squatting, bending to the left or right, and standing still. This is a robust solution for developing a yoga, dance, metaverse, and fitness application that checks for the correction of the pose without having any additional monitor like a personal trainer. MediaPipe solution offers an open-source cross-platform which utilizes a two-step detector-tracker ML pipeline for live detection of key landmarks on our body which can be used for motion data collection. The prediction of real-time poses uses a variety of machine learning techniques and different types of analysis. Without primarily relying on powerful desktop environments for inference, our method achieves real-time performance on the majority of contemporary mobile phones, desktops/laptops, Python, or even the web. Experimental results show that our method outperforms the existing method in terms of accuracy and real-time capability, achieving an accuracy of 99.92% on testing datasets.

Keywords: human activity detection, media pipe, machine learning, metaverse applications

Procedia PDF Downloads 163
4592 Damage Detection in a Cantilever Beam under Different Excitation and Temperature Conditions

Authors: A. Kyprianou, A. Tjirkallis

Abstract:

Condition monitoring of structures in service is very important as it provides information about the risk of damage development. One of the essential constituents of structural condition monitoring is the damage detection methodology. In the context of condition monitoring of in service structures a damage detection methodology analyses data obtained from the structure while it is in operation. Usually, this means that the data could be affected by operational and environmental conditions in a way that could mask the effects of a possible damage on the data. This, depending on the damage detection methodology, could lead to either false alarms or miss existing damages. In this article a damage detection methodology that is based on the Spatio-temporal continuous wavelet transform (SPT-CWT) analysis of a sequence of experimental time responses of a cantilever beam is proposed. The cantilever is subjected to white and pink noise excitation to simulate different operating conditions. In addition, in order to simulate changing environmental conditions, the cantilever is subjected to heating by a heat gun. The response of the cantilever beam is measured by a high-speed camera. Edges are extracted from the series of images of the beam response captured by the camera. Subsequent processing of the edges gives a series of time responses on 439 points on the beam. This sequence is then analyzed using the SPT-CWT to identify damage. The algorithm proposed was able to clearly identify damage under any condition when the structure was excited by white noise force. In addition, in the case of white noise excitation, the analysis could also reveal the position of the heat gun when it was used to heat the structure. The analysis could identify the different operating conditions i.e. between responses due to white noise excitation and responses due to pink noise excitation. During the pink noise excitation whereas damage and changing temperature were identified it was not possible to clearly identify the effect of damage from that of temperature. The methodology proposed in this article for damage detection enables the separation the damage effect from that due to temperature and excitation on data obtained from measurements of a cantilever beam. This methodology does not require information about the apriori state of the structure.

Keywords: spatiotemporal continuous wavelet transform, damage detection, data normalization, varying temperature

Procedia PDF Downloads 268
4591 Detection and Classification of Rubber Tree Leaf Diseases Using Machine Learning

Authors: Kavyadevi N., Kaviya G., Gowsalya P., Janani M., Mohanraj S.

Abstract:

Hevea brasiliensis, also known as the rubber tree, is one of the foremost assets of crops in the world. One of the most significant advantages of the Rubber Plant in terms of air oxygenation is its capacity to reduce the likelihood of an individual developing respiratory allergies like asthma. To construct such a system that can properly identify crop diseases and pests and then create a database of insecticides for each pest and disease, we must first give treatment for the illness that has been detected. We shall primarily examine three major leaf diseases since they are economically deficient in this article, which is Bird's eye spot, algal spot and powdery mildew. And the recommended work focuses on disease identification on rubber tree leaves. It will be accomplished by employing one of the superior algorithms. Input, Preprocessing, Image Segmentation, Extraction Feature, and Classification will be followed by the processing technique. We will use time-consuming procedures that they use to detect the sickness. As a consequence, the main ailments, underlying causes, and signs and symptoms of diseases that harm the rubber tree are covered in this study.

Keywords: image processing, python, convolution neural network (CNN), machine learning

Procedia PDF Downloads 64
4590 Research on Placement Method of the Magnetic Flux Leakage Sensor Based on Online Detection of the Transformer Winding Deformation

Authors: Wei Zheng, Mao Ji, Zhe Hou, Meng Huang, Bo Qi

Abstract:

The transformer is the key equipment of the power system. Winding deformation is one of the main transformer defects, and timely and effective detection of the transformer winding deformation can ensure the safe and stable operation of the transformer to the maximum extent. When winding deformation occurs, the size, shape and spatial position of the winding will change, which directly leads to the change of magnetic flux leakage distribution. Therefore, it is promising to study the online detection method of the transformer winding deformation based on magnetic flux leakage characteristics, in which the key step is to study the optimal placement method of magnetic flux leakage sensors inside the transformer. In this paper, a simulation model of the transformer winding deformation is established to obtain the internal magnetic flux leakage distribution of the transformer under normal operation and different winding deformation conditions, and the law of change of magnetic flux leakage distribution due to winding deformation is analyzed. The results show that different winding deformation leads to different characteristics of the magnetic flux leakage distribution. On this basis, an optimized placement of magnetic flux leakage sensors inside the transformer is proposed to provide a basis for the online detection method of transformer winding deformation based on the magnetic flux leakage characteristics.

Keywords: magnetic flux leakage, sensor placement method, transformer, winding deformation

Procedia PDF Downloads 178
4589 Model Updating-Based Approach for Damage Prognosis in Frames via Modal Residual Force

Authors: Gholamreza Ghodrati Amiri, Mojtaba Jafarian Abyaneh, Ali Zare Hosseinzadeh

Abstract:

This paper presents an effective model updating strategy for damage localization and quantification in frames by defining damage detection problem as an optimization issue. A generalized version of the Modal Residual Force (MRF) is employed for presenting a new damage-sensitive cost function. Then, Grey Wolf Optimization (GWO) algorithm is utilized for solving suggested inverse problem and the global extremums are reported as damage detection results. The applicability of the presented method is investigated by studying different damage patterns on the benchmark problem of the IASC-ASCE, as well as a planar shear frame structure. The obtained results emphasize good performance of the method not only in free-noise cases, but also when the input data are contaminated with different levels of noises.

Keywords: frame, grey wolf optimization algorithm, modal residual force, structural damage detection

Procedia PDF Downloads 372
4588 Using Scale Invariant Feature Transform Features to Recognize Characters in Natural Scene Images

Authors: Belaynesh Chekol, Numan Çelebi

Abstract:

The main purpose of this work is to recognize individual characters extracted from natural scene images using scale invariant feature transform (SIFT) features as an input to K-nearest neighbor (KNN); a classification learner algorithm. For this task, 1,068 and 78 images of English alphabet characters taken from Chars74k data set is used to train and test the classifier respectively. For each character image, We have generated describing features by using SIFT algorithm. This set of features is fed to the learner so that it can recognize and label new images of English characters. Two types of KNN (fine KNN and weighted KNN) were trained and the resulted classification accuracy is 56.9% and 56.5% respectively. The training time taken was the same for both fine and weighted KNN.

Keywords: character recognition, KNN, natural scene image, SIFT

Procedia PDF Downloads 272
4587 Optimized Parameters for Simultaneous Detection of Cd²⁺, Pb²⁺ and CO²⁺ Ions in Water Using Square Wave Voltammetry on the Unmodified Glassy Carbon Electrode

Authors: K. Sruthi, Sai Snehitha Yadavalli, Swathi Gosh Acharyya

Abstract:

Water is the most crucial element for sustaining life on earth. Increasing water pollution directly or indirectly leads to harmful effects on human life. Most of the heavy metal ions are harmful in their cationic form. These heavy metal ions are released by various activities like disposing of batteries, industrial wastes, automobile emissions, and soil contamination. Ions like (Pb, Co, Cd) are carcinogenic and show many harmful effects when consumed more than certain limits proposed by WHO. The simultaneous detection of the heavy metal ions (Pb, Co, Cd), which are highly toxic, is reported in this study. There are many analytical methods for quantifying, but electrochemical techniques are given high priority because of their sensitivity and ability to detect and recognize lower concentrations. Square wave voltammetry was preferred in electrochemical methods due to the absence of background currents which is interference. Square wave voltammetry was performed on GCE for the quantitative detection of ions. Three electrode system consisting of a glassy carbon electrode as the working electrode (3 mm diameter), Ag/Agcl electrode as the reference electrode, and a platinum wire as the counter electrode was chosen for experimentation. The mechanism of detection was done by optimizing the experimental parameters, namely pH, scan rate, and temperature. Under the optimized conditions, square wave voltammetry was performed for simultaneous detection. Scan rates were varied from 5 mV/s to 100 mV/s and found that at 25 mV/s all the three ions were detected simultaneously with proper peaks at particular stripping potential. The variation of pH from 3 to 8 was done where the optimized pH was taken as pH 5 which holds good for three ions. There was a decreasing trend at starting because of hydrogen gas evolution, and after pH 5 again there was a decreasing trend that is because of hydroxide formation on the surface of the working electrode (GCE). The temperature variation from 25˚C to 45˚C was done where the optimum temperature concerning three ions was taken as 35˚C. Deposition and stripping potentials were given as +1.5 V and -1.5 V, and the resting time of 150 seconds was given. Three ions were detected at stripping potentials of Cd²⁺ at -0.84 V, Pb²⁺ at -0.54 V, and Co²⁺ at -0.44 V. The parameters of detection were optimized on a glassy carbon electrode for simultaneous detection of the ions at lower concentrations by square wave voltammetry.

Keywords: cadmium, cobalt, lead, glassy carbon electrode, square wave anodic stripping voltammetry

Procedia PDF Downloads 100
4586 Machine Learning and Deep Learning Approach for People Recognition and Tracking in Crowd for Safety Monitoring

Authors: A. Degale Desta, Cheng Jian

Abstract:

Deep learning application in computer vision is rapidly advancing, giving it the ability to monitor the public and quickly identify potentially anomalous behaviour from crowd scenes. Therefore, the purpose of the current work is to improve the performance of safety of people in crowd events from panic behaviour through introducing the innovative idea of Aggregation of Ensembles (AOE), which makes use of the pre-trained ConvNets and a pool of classifiers to find anomalies in video data with packed scenes. According to the theory of algorithms that applied K-means, KNN, CNN, SVD, and Faster-CNN, YOLOv5 architectures learn different levels of semantic representation from crowd videos; the proposed approach leverages an ensemble of various fine-tuned convolutional neural networks (CNN), allowing for the extraction of enriched feature sets. In addition to the above algorithms, a long short-term memory neural network to forecast future feature values and a handmade feature that takes into consideration the peculiarities of the crowd to understand human behavior. On well-known datasets of panic situations, experiments are run to assess the effectiveness and precision of the suggested method. Results reveal that, compared to state-of-the-art methodologies, the system produces better and more promising results in terms of accuracy and processing speed.

Keywords: action recognition, computer vision, crowd detecting and tracking, deep learning

Procedia PDF Downloads 147
4585 A Deep Learning Approach to Subsection Identification in Electronic Health Records

Authors: Nitin Shravan, Sudarsun Santhiappan, B. Sivaselvan

Abstract:

Subsection identification, in the context of Electronic Health Records (EHRs), is identifying the important sections for down-stream tasks like auto-coding. In this work, we classify the text present in EHRs according to their information, using machine learning and deep learning techniques. We initially describe briefly about the problem and formulate it as a text classification problem. Then, we discuss upon the methods from the literature. We try two approaches - traditional feature extraction based machine learning methods and deep learning methods. Through experiments on a private dataset, we establish that the deep learning methods perform better than the feature extraction based Machine Learning Models.

Keywords: deep learning, machine learning, semantic clinical classification, subsection identification, text classification

Procedia PDF Downloads 201
4584 Robust and Real-Time Traffic Counting System

Authors: Hossam M. Moftah, Aboul Ella Hassanien

Abstract:

In the recent years the importance of automatic traffic control has increased due to the traffic jams problem especially in big cities for signal control and efficient traffic management. Traffic counting as a kind of traffic control is important to know the road traffic density in real time. This paper presents a fast and robust traffic counting system using different image processing techniques. The proposed system is composed of the following four fundamental building phases: image acquisition, pre-processing, object detection, and finally counting the connected objects. The object detection phase is comprised of the following five steps: subtracting the background, converting the image to binary, closing gaps and connecting nearby blobs, image smoothing to remove noises and very small objects, and detecting the connected objects. Experimental results show the great success of the proposed approach.

Keywords: traffic counting, traffic management, image processing, object detection, computer vision

Procedia PDF Downloads 283
4583 EZOB Technology, Biomass Gasification, and Microcogeneration Unit

Authors: Martin Lisý, Marek Baláš, Michal Špiláček, Zdeněk Skála

Abstract:

This paper deals with the issue of biomass and sorted municipal waste gasification and cogeneration using hot air turbo set. It brings description of designed pilot plant with electrical output 80 kWe. The generated gas is burned in secondary combustion chamber located beyond the gas generator. Flue gas flows through the heat exchanger where the compressed air is heated and consequently brought to a micro turbine. Except description, this paper brings our basic experiences from operating of pilot plant (operating parameters, contributions, problems during operating, etc.). The principal advantage of the given cycle is the fact that there is no contact between the generated gas and the turbine. So there is no need for costly and complicated gas cleaning which is the main source of operating problems in direct use in combustion engines because the content of impurities in the gas causes operation problems to the units due to clogging and tarring of working surfaces of engines and turbines, which may lead as far as serious damage to the equipment under operation. Another merit is the compact container package making installation of the facility easier or making it relatively more mobile. We imagine, this solution of cogeneration from biomass or waste can be suitable for small industrial or communal applications, for low output cogeneration.

Keywords: biomass, combustion, gasification, microcogeneration

Procedia PDF Downloads 324
4582 Biomass Gasification and Microcogeneration Unit–EZOB Technology

Authors: Martin Lisý, Marek Baláš, Michal Špiláček, Zdeněk Skála

Abstract:

This paper deals with the issue of biomass and sorted municipal waste gasification and cogeneration using hot-air turbo-set. It brings description of designed pilot plant with electrical output 80 kWe. The generated gas is burned in secondary combustion chamber located beyond the gas generator. Flue gas flows through the heat exchanger where the compressed air is heated and consequently brought to a micro turbine. Except description, this paper brings our basic experiences from operating of pilot plant (operating parameters, contributions, problems during operating, etc.). The principal advantage of the given cycle is the fact that there is no contact between the generated gas and the turbine. So there is no need for costly and complicated gas cleaning which is the main source of operating problems in direct use in combustion engines because the content of impurities in the gas causes operation problems to the units due to clogging and tarring of working surfaces of engines and turbines, which may lead as far as serious damage to the equipment under operation. Another merit is the compact container package making installation of the facility easier or making it relatively more mobile. We imagine, this solution of cogeneration from biomass or waste can be suitable for small industrial or communal applications, for low output cogeneration.

Keywords: biomass, combustion, gasification, microcogeneration

Procedia PDF Downloads 480
4581 1-D Convolutional Neural Network Approach for Wheel Flat Detection for Freight Wagons

Authors: Dachuan Shi, M. Hecht, Y. Ye

Abstract:

With the trend of digitalization in railway freight transport, a large number of freight wagons in Germany have been equipped with telematics devices, commonly placed on the wagon body. A telematics device contains a GPS module for tracking and a 3-axis accelerometer for shock detection. Besides these basic functions, it is desired to use the integrated accelerometer for condition monitoring without any additional sensors. Wheel flats as a common type of failure on wheel tread cause large impacts on wagons and infrastructure as well as impulsive noise. A large wheel flat may even cause safety issues such as derailments. In this sense, this paper proposes a machine learning approach for wheel flat detection by using car body accelerations. Due to suspension systems, impulsive signals caused by wheel flats are damped significantly and thus could be buried in signal noise and disturbances. Therefore, it is very challenging to detect wheel flats using car body accelerations. The proposed algorithm considers the envelope spectrum of car body accelerations to eliminate the effect of noise and disturbances. Subsequently, a 1-D convolutional neural network (CNN), which is well known as a deep learning method, is constructed to automatically extract features in the envelope-frequency domain and conduct classification. The constructed CNN is trained and tested on field test data, which are measured on the underframe of a tank wagon with a wheel flat of 20 mm length in the operational condition. The test results demonstrate the good performance of the proposed algorithm for real-time fault detection.

Keywords: fault detection, wheel flat, convolutional neural network, machine learning

Procedia PDF Downloads 118
4580 A DNA-Based Nano-biosensor for the Rapid Detection of the Dengue Virus in Mosquito

Authors: Lilia M. Fernando, Matthew K. Vasher, Evangelyn C. Alocilja

Abstract:

This paper describes the development of a DNA-based nanobiosensor to detect the dengue virus in mosquito using electrically active magnetic (EAM) nanoparticles as the concentrator and electrochemical transducer. The biosensor detection encompasses two sets of oligonucleotide probes that are specific to the dengue virus: the detector probe labeled with the EAM nanoparticles and the biotinylated capture probe. The DNA targets are double hybridized to the detector and the capture probes and concentrated from nonspecific DNA fragments by applying a magnetic field. Subsequently, the DNA sandwiched targets (EAM-detector probe–DNA target–capture probe-biotin) are captured on streptavidin modified screen printed carbon electrodes through the biotinylated capture probes. Detection is achieved electrochemically by measuring the oxidation–reduction signal of the EAM nanoparticles. Results indicate that the biosensor is able to detect the redox signal of the EAM nanoparticles at dengue DNA concentrations as low as 10 ng/ul.

Keywords: dengue, magnetic nanoparticles, mosquito, nanobiosensor

Procedia PDF Downloads 352
4579 Detection of Micro-Unmanned Ariel Vehicles Using a Multiple-Input Multiple-Output Digital Array Radar

Authors: Tareq AlNuaim, Mubashir Alam, Abdulrazaq Aldowesh

Abstract:

The usage of micro-Unmanned Ariel Vehicles (UAVs) has witnessed an enormous increase recently. Detection of such drones became a necessity nowadays to prevent any harmful activities. Typically, such targets have low velocity and low Radar Cross Section (RCS), making them indistinguishable from clutter and phase noise. Multiple-Input Multiple-Output (MIMO) Radars have many potentials; it increases the degrees of freedom on both transmit and receive ends. Such architecture allows for flexibility in operation, through utilizing the direct access to every element in the transmit/ receive array. MIMO systems allow for several array processing techniques, permitting the system to stare at targets for longer times, which improves the Doppler resolution. In this paper, a 2×2 MIMO radar prototype is developed using Software Defined Radio (SDR) technology, and its performance is evaluated against a slow-moving low radar cross section micro-UAV used by hobbyists. Radar cross section simulations were carried out using FEKO simulator, achieving an average of -14.42 dBsm at S-band. The developed prototype was experimentally evaluated achieving more than 300 meters of detection range for a DJI Mavic pro-drone

Keywords: digital beamforming, drone detection, micro-UAV, MIMO, phased array

Procedia PDF Downloads 126
4578 Comparison of Direction of Arrival Estimation Method for Drone Based on Phased Microphone Array

Authors: Jiwon Lee, Yeong-Ju Go, Jong-Soo Choi

Abstract:

Drones were first developed for military use and were used in World War 1. But recently drones have been used in a variety of fields. Several companies actively utilize drone technology to strengthen their services, and in agriculture, drones are used for crop monitoring and sowing. Other people use drones for hobby activities such as photography. However, as the range of use of drones expands rapidly, problems caused by drones such as improperly flying, privacy and terrorism are also increasing. As the need for monitoring and tracking of drones increases, researches are progressing accordingly. The drone detection system estimates the position of the drone using the physical phenomena that occur when the drones fly. The drone detection system measures being developed utilize many approaches, such as radar, infrared camera, and acoustic detection systems. Among the various drone detection system, the acoustic detection system is advantageous in that the microphone array system is small, inexpensive, and easy to operate than other systems. In this paper, the acoustic signal is acquired by using minimum microphone when drone is flying, and direction of drone is estimated. When estimating the Direction of Arrival(DOA), there is a method of calculating the DOA based on the Time Difference of Arrival(TDOA) and a method of calculating the DOA based on the beamforming. The TDOA technique requires less number of microphones than the beamforming technique, but is weak in noisy environments and can only estimate the DOA of a single source. The beamforming technique requires more microphones than the TDOA technique. However, it is strong against the noisy environment and it is possible to simultaneously estimate the DOA of several drones. When estimating the DOA using acoustic signals emitted from the drone, it is impossible to measure the position of the drone, and only the direction can be estimated. To overcome this problem, in this work we show how to estimate the position of drones by arranging multiple microphone arrays. The microphone array used in the experiments was four tetrahedral microphones. We simulated the performance of each DOA algorithm and demonstrated the simulation results through experiments.

Keywords: acoustic sensing, direction of arrival, drone detection, microphone array

Procedia PDF Downloads 146
4577 Contactless Attendance System along with Temperature Monitoring

Authors: Nalini C. Iyer, Shraddha H., Anagha B. Varahamurthy, Dikshith C. S., Ishwar G. Kubasad, Vinayak I. Karalatti, Pavan B. Mulimani

Abstract:

The current scenario of the pandemic due to COVID-19 has led to the awareness among the people to avoid unneces-sary contact in public places. There is a need to avoid contact with physical objects to stop the spreading of infection. The contactless feature has to be included in the systems in public places wherever possible. For example, attendance monitoring systems with fingerprint biometric can be replaced with a contactless feature. One more important protocol followed in the current situation is temperature monitoring and screening. The paper describes an attendance system with a contactless feature and temperature screening for the university. The system displays a QR code to scan, which redirects to the student login web page only if the location is valid (the location where the student scans the QR code should be the location of the display of the QR code). Once the student logs in, the temperature of the student is scanned by the contactless temperature sensor (mlx90614) with an error of 0.5°C. If the temperature falls in the range of the desired value (range of normal body temperature), then the attendance of the student is marked as present, stored in the database, and the door opens automatically. The attendance is marked as absent in the other case, alerted with the display of temperature, and the door remains closed. The door is automated with the help of a servomotor. To avoid the proxy, IR sensors are used to count the number of students in the classroom. The hardware system consisting of a contactless temperature sensor and IR sensor is implemented on the microcontroller, NodeMCU.

Keywords: NodeMCU, IR sensor, attendance monitoring, contactless, temperature

Procedia PDF Downloads 176
4576 Biosensor Technologies in Neurotransmitters Detection

Authors: Joanna Cabaj, Sylwia Baluta, Karol Malecha

Abstract:

Catecholamines are vital neurotransmitters that mediate a variety of central nervous system functions, such as motor control, cognition, emotion, memory processing, and endocrine modulation. Dysfunctions in catecholamine neurotransmission are induced in some neurologic and neuropsychiatric diseases. Changeable neurotransmitters level in biological fluids can be a marker of several neurological disorders. Because of its significance in analytical techniques and diagnostics, sensitive and selective detection of neurotransmitters is increasingly attracting a lot of attention in different areas of bio-analysis or biomedical research. Recently, optical techniques for the detection of catecholamines have attracted interests due to their reasonable cost, convenient control, as well as maneuverability in biological environments. Nevertheless, with the observed need for a sensitive and selective catecholamines sensor, the development of a convenient method for this neurotransmitter is still at its basic level. The manipulation of nanostructured materials in conjunction with biological molecules has led to the development of a new class of hybrid-modified enzymatic sensors in which both enhancement of charge transport and biological activity preservation may be obtained. Immobilization of biomaterials on electrode surfaces is the crucial step in fabricating electrochemical as well as optical biosensors and bioelectronic devices. Continuing systematic investigation in manufacturing of enzyme–conducting sensitive systems, here is presented a convenient fluorescence as well as electrochemical sensing strategy for catecholamines detection.

Keywords: biosensors, catecholamines, fluorescence, enzymes

Procedia PDF Downloads 100
4575 USE-Net: SE-Block Enhanced U-Net Architecture for Robust Speaker Identification

Authors: Kilari Nikhil, Ankur Tibrewal, Srinivas Kruthiventi S. S.

Abstract:

Conventional speaker identification systems often fall short of capturing the diverse variations present in speech data due to fixed-scale architectures. In this research, we propose a CNN-based architecture, USENet, designed to overcome these limitations. Leveraging two key techniques, our approach achieves superior performance on the VoxCeleb 1 Dataset without any pre-training. Firstly, we adopt a U-net-inspired design to extract features at multiple scales, empowering our model to capture speech characteristics effectively. Secondly, we introduce the squeeze and excitation block to enhance spatial feature learning. The proposed architecture showcases significant advancements in speaker identification, outperforming existing methods, and holds promise for future research in this domain.

Keywords: multi-scale feature extraction, squeeze and excitation, VoxCeleb1 speaker identification, mel-spectrograms, USENet

Procedia PDF Downloads 61
4574 Application on Metastable Measurement with Wide Range High Resolution VDL Circuit

Authors: Po-Hui Yang, Jing-Min Chen, Po-Yu Kuo, Chia-Chun Wu

Abstract:

This paper proposed a high resolution Vernier Delay Line (VDL) measurement circuit with coarse and fine detection mechanism, which improved the trade-off problem between high resolution and less delay cells in traditional VDL circuits. And the measuring time of proposed measurement circuit is also under the high resolution requests. At first, the testing range of input signal which proposed high resolution delay line is detected by coarse detection VDL. Moreover, the delayed input signal is transmitted to fine detection VDL for measuring value with better accuracy. This paper is implemented at 0.18μm process, operating frequency is 100 MHz, and the resolution achieved 2.0 ps with only 16-stage delay cells. The test range is 170ps wide, and 17% stages saved compare with traditional single delay line circuit.

Keywords: vernier delay line, D-type flip-flop, DFF, metastable phenomenon

Procedia PDF Downloads 587
4573 Registration of Multi-Temporal Unmanned Aerial Vehicle Images for Facility Monitoring

Authors: Dongyeob Han, Jungwon Huh, Quang Huy Tran, Choonghyun Kang

Abstract:

Unmanned Aerial Vehicles (UAVs) have been used for surveillance, monitoring, inspection, and mapping. In this paper, we present a systematic approach for automatic registration of UAV images for monitoring facilities such as building, green house, and civil structures. The two-step process is applied; 1) an image matching technique based on SURF (Speeded up Robust Feature) and RANSAC (Random Sample Consensus), 2) bundle adjustment of multi-temporal images. Image matching to find corresponding points is one of the most important steps for the precise registration of multi-temporal images. We used the SURF algorithm to find a quick and effective matching points. RANSAC algorithm was used in the process of finding matching points between images and in the bundle adjustment process. Experimental results from UAV images showed that our approach has a good accuracy to be applied to the change detection of facility.

Keywords: building, image matching, temperature, unmanned aerial vehicle

Procedia PDF Downloads 281
4572 Improve Divers Tracking and Classification in Sonar Images Using Robust Diver Wake Detection Algorithm

Authors: Mohammad Tarek Al Muallim, Ozhan Duzenli, Ceyhun Ilguy

Abstract:

Harbor protection systems are so important. The need for automatic protection systems has increased over the last years. Diver detection active sonar has great significance. It used to detect underwater threats such as divers and autonomous underwater vehicle. To automatically detect such threats the sonar image is processed by algorithms. These algorithms used to detect, track and classify of underwater objects. In this work, divers tracking and classification algorithm is improved be proposing a robust wake detection method. To detect objects the sonar images is normalized then segmented based on fixed threshold. Next, the centroids of the segments are found and clustered based on distance metric. Then to track the objects linear Kalman filter is applied. To reduce effect of noise and creation of false tracks, the Kalman tracker is fine tuned. The tuning is done based on our active sonar specifications. After the tracks are initialed and updated they are subjected to a filtering stage to eliminate the noisy and unstable tracks. Also to eliminate object with a speed out of the diver speed range such as buoys and fast boats. Afterwards the result tracks are subjected to a classification stage to deiced the type of the object been tracked. Here the classification stage is to deice wither if the tracked object is an open circuit diver or a close circuit diver. At the classification stage, a small area around the object is extracted and a novel wake detection method is applied. The morphological features of the object with his wake is extracted. We used support vector machine to find the best classifier. The sonar training images and the test images are collected by ARMELSAN Defense Technologies Company using the portable diver detection sonar ARAS-2023. After applying the algorithm to the test sonar data, we get fine and stable tracks of the divers. The total classification accuracy achieved with the diver type is 97%.

Keywords: harbor protection, diver detection, active sonar, wake detection, diver classification

Procedia PDF Downloads 226
4571 A Real-Time Moving Object Detection and Tracking Scheme and Its Implementation for Video Surveillance System

Authors: Mulugeta K. Tefera, Xiaolong Yang, Jian Liu

Abstract:

Detection and tracking of moving objects are very important in many application contexts such as detection and recognition of people, visual surveillance and automatic generation of video effect and so on. However, the task of detecting a real shape of an object in motion becomes tricky due to various challenges like dynamic scene changes, presence of shadow, and illumination variations due to light switch. For such systems, once the moving object is detected, tracking is also a crucial step for those applications that used in military defense, video surveillance, human computer interaction, and medical diagnostics as well as in commercial fields such as video games. In this paper, an object presents in dynamic background is detected using adaptive mixture of Gaussian based analysis of the video sequences. Then the detected moving object is tracked using the region based moving object tracking and inter-frame differential mechanisms to address the partial overlapping and occlusion problems. Firstly, the detection algorithm effectively detects and extracts the moving object target by enhancing and post processing morphological operations. Secondly, the extracted object uses region based moving object tracking and inter-frame difference to improve the tracking speed of real-time moving objects in different video frames. Finally, the plotting method was applied to detect the moving objects effectively and describes the object’s motion being tracked. The experiment has been performed on image sequences acquired both indoor and outdoor environments and one stationary and web camera has been used.

Keywords: background modeling, Gaussian mixture model, inter-frame difference, object detection and tracking, video surveillance

Procedia PDF Downloads 463
4570 Integration of Educational Data Mining Models to a Web-Based Support System for Predicting High School Student Performance

Authors: Sokkhey Phauk, Takeo Okazaki

Abstract:

The challenging task in educational institutions is to maximize the high performance of students and minimize the failure rate of poor-performing students. An effective method to leverage this task is to know student learning patterns with highly influencing factors and get an early prediction of student learning outcomes at the timely stage for setting up policies for improvement. Educational data mining (EDM) is an emerging disciplinary field of data mining, statistics, and machine learning concerned with extracting useful knowledge and information for the sake of improvement and development in the education environment. The study is of this work is to propose techniques in EDM and integrate it into a web-based system for predicting poor-performing students. A comparative study of prediction models is conducted. Subsequently, high performing models are developed to get higher performance. The hybrid random forest (Hybrid RF) produces the most successful classification. For the context of intervention and improving the learning outcomes, a feature selection method MICHI, which is the combination of mutual information (MI) and chi-square (CHI) algorithms based on the ranked feature scores, is introduced to select a dominant feature set that improves the performance of prediction and uses the obtained dominant set as information for intervention. By using the proposed techniques of EDM, an academic performance prediction system (APPS) is subsequently developed for educational stockholders to get an early prediction of student learning outcomes for timely intervention. Experimental outcomes and evaluation surveys report the effectiveness and usefulness of the developed system. The system is used to help educational stakeholders and related individuals for intervening and improving student performance.

Keywords: academic performance prediction system, educational data mining, dominant factors, feature selection method, prediction model, student performance

Procedia PDF Downloads 97
4569 Unsupervised Feature Learning by Pre-Route Simulation of Auto-Encoder Behavior Model

Authors: Youngjae Jin, Daeshik Kim

Abstract:

This paper describes a cycle accurate simulation results of weight values learned by an auto-encoder behavior model in terms of pre-route simulation. Given the results we visualized the first layer representations with natural images. Many common deep learning threads have focused on learning high-level abstraction of unlabeled raw data by unsupervised feature learning. However, in the process of handling such a huge amount of data, the learning method’s computation complexity and time limited advanced research. These limitations came from the fact these algorithms were computed by using only single core CPUs. For this reason, parallel-based hardware, FPGAs, was seen as a possible solution to overcome these limitations. We adopted and simulated the ready-made auto-encoder to design a behavior model in Verilog HDL before designing hardware. With the auto-encoder behavior model pre-route simulation, we obtained the cycle accurate results of the parameter of each hidden layer by using MODELSIM. The cycle accurate results are very important factor in designing a parallel-based digital hardware. Finally this paper shows an appropriate operation of behavior model based pre-route simulation. Moreover, we visualized learning latent representations of the first hidden layer with Kyoto natural image dataset.

Keywords: auto-encoder, behavior model simulation, digital hardware design, pre-route simulation, Unsupervised feature learning

Procedia PDF Downloads 431
4568 Graph Neural Networks and Rotary Position Embedding for Voice Activity Detection

Authors: YingWei Tan, XueFeng Ding

Abstract:

Attention-based voice activity detection models have gained significant attention in recent years due to their fast training speed and ability to capture a wide contextual range. The inclusion of multi-head style and position embedding in the attention architecture are crucial. Having multiple attention heads allows for differential focus on different parts of the sequence, while position embedding provides guidance for modeling dependencies between elements at various positions in the input sequence. In this work, we propose an approach by considering each head as a node, enabling the application of graph neural networks (GNN) to identify correlations among the different nodes. In addition, we adopt an implementation named rotary position embedding (RoPE), which encodes absolute positional information into the input sequence by a rotation matrix, and naturally incorporates explicit relative position information into a self-attention module. We evaluate the effectiveness of our method on a synthetic dataset, and the results demonstrate its superiority over the baseline CRNN in scenarios with low signal-to-noise ratio and noise, while also exhibiting robustness across different noise types. In summary, our proposed framework effectively combines the strengths of CNN and RNN (LSTM), and further enhances detection performance through the integration of graph neural networks and rotary position embedding.

Keywords: voice activity detection, CRNN, graph neural networks, rotary position embedding

Procedia PDF Downloads 53
4567 A Tool to Measure Efficiency and Trust Towards eXplainable Artificial Intelligence in Conflict Detection Tasks

Authors: Raphael Tuor, Denis Lalanne

Abstract:

The ATM research community is missing suitable tools to design, test, and validate new UI prototypes. Important stakes underline the implementation of both DSS and XAI methods into current systems. ML-based DSS are gaining in relevance as ATFM becomes increasingly complex. However, these systems only prove useful if a human can understand them, and thus new XAI methods are needed. The human-machine dyad should work as a team and should understand each other. We present xSky, a configurable benchmark tool that allows us to compare different versions of an ATC interface in conflict detection tasks. Our main contributions to the ATC research community are (1) a conflict detection task simulator (xSky) that allows to test the applicability of visual prototypes on scenarios of varying difficulty and outputting relevant operational metrics (2) a theoretical approach to the explanations of AI-driven trajectory predictions. xSky addresses several issues that were identified within available research tools. Researchers can configure the dimensions affecting scenario difficulty with a simple CSV file. Both the content and appearance of the XAI elements can be customized in a few steps. As a proof-of-concept, we implemented an XAI prototype inspired by the maritime field.

Keywords: air traffic control, air traffic simulation, conflict detection, explainable artificial intelligence, explainability, human-automation collaboration, human factors, information visualization, interpretability, trajectory prediction

Procedia PDF Downloads 150
4566 Agent-Base Modeling of IoT Applications by Using Software Product Line

Authors: Asad Abbas, Muhammad Fezan Afzal, Muhammad Latif Anjum, Muhammad Azmat

Abstract:

The Internet of Things (IoT) is used to link up real objects that use the internet to interact. IoT applications allow handling and operating the equipment in accordance with environmental needs, such as transportation and healthcare. IoT devices are linked together via a number of agents that act as a middleman for communications. The operation of a heat sensor differs indoors and outside because agent applications work with environmental variables. In this article, we suggest using Software Product Line (SPL) to model IoT agents and applications' features on an XML-based basis. The contextual diversity within the same domain of application can be handled, and the reusability of features is increased by XML-based feature modelling. For the purpose of managing contextual variability, we have embraced XML for modelling IoT applications, agents, and internet-connected devices.

Keywords: IoT agents, IoT applications, software product line, feature model, XML

Procedia PDF Downloads 80
4565 Detection of Pharmaceutical Personal Protective Equipment in Video Stream

Authors: Michael Leontiev, Danil Zhilikov, Dmitry Lobanov, Lenar Klimov, Vyacheslav Chertan, Daniel Bobrov, Vladislav Maslov, Vasilii Vologdin, Ksenia Balabaeva

Abstract:

Pharmaceutical manufacturing is a complex process, where each stage requires a high level of safety and sterility. Personal Protective Equipment (PPE) is used for this purpose. Despite all the measures of control, the human factor (improper PPE wearing) causes numerous losses to human health and material property. This research proposes a solid computer vision system for ensuring safety in pharmaceutical laboratories. For this, we have tested a wide range of state-of-the-art object detection methods. Composing previously obtained results in this sphere with our own approach to this problem, we have reached a high accuracy ([email protected]) ranging from 0.77 up to 0.98 in detecting all the elements of a common set of PPE used in pharmaceutical laboratories. Our system is a step towards safe medicine production.

Keywords: sterility and safety in pharmaceutical development, personal protective equipment, computer vision, object detection, monitoring in pharmaceutical development, PPE

Procedia PDF Downloads 65
4564 Detection of Safety Goggles on Humans in Industrial Environment Using Faster-Region Based on Convolutional Neural Network with Rotated Bounding Box

Authors: Ankit Kamboj, Shikha Talwar, Nilesh Powar

Abstract:

To successfully deliver our products in the market, the employees need to be in a safe environment, especially in an industrial and manufacturing environment. The consequences of delinquency in wearing safety glasses while working in industrial plants could be high risk to employees, hence the need to develop a real-time automatic detection system which detects the persons (violators) not wearing safety glasses. In this study a convolutional neural network (CNN) algorithm called faster region based CNN (Faster RCNN) with rotated bounding box has been used for detecting safety glasses on persons; the algorithm has an advantage of detecting safety glasses with different orientation angles on the persons. The proposed method of rotational bounding boxes with a convolutional neural network first detects a person from the images, and then the method detects whether the person is wearing safety glasses or not. The video data is captured at the entrance of restricted zones of the industrial environment (manufacturing plant), which is further converted into images at 2 frames per second. In the first step, the CNN with pre-trained weights on COCO dataset is used for person detection where the detections are cropped as images. Then the safety goggles are labelled on the cropped images using the image labelling tool called roLabelImg, which is used to annotate the ground truth values of rotated objects more accurately, and the annotations obtained are further modified to depict four coordinates of the rectangular bounding box. Next, the faster RCNN with rotated bounding box is used to detect safety goggles, which is then compared with traditional bounding box faster RCNN in terms of detection accuracy (average precision), which shows the effectiveness of the proposed method for detection of rotatory objects. The deep learning benchmarking is done on a Dell workstation with a 16GB Nvidia GPU.

Keywords: CNN, deep learning, faster RCNN, roLabelImg rotated bounding box, safety goggle detection

Procedia PDF Downloads 121