JOIV : International Journal on Informatics Visualization
Not a member yet
786 research outputs found
Sort by
Color and Attention for U : Modified Multi Attention U-Net for a Better Image Colorization
Get PDFImage colorization is a tedious task that requires creativity and understanding of the image context and semantic information. Many models have been made by harnessing various deep learning architectures to learn the plausible colorization. With the rapid discovery of new architecture and image generation techniques, more powerful options can be explored and improved for image colorization tasks. This research explores a new architecture to colorize an image by using pre-trained embeddings on U-Net combined with several attention modules across the model. Using embeddings from a pre-trained classifier provides a high-level feature extraction from the image. Conversely, multi-attention gives a little taste of image segmentation so that the model can distinguish objects in the image and further enhance the additional information given by the pre-trained embeddings. Adversarial training is also utilized as a normalization to make the generated image more realistic. This research preferred Parch GAN over base GAN as the discriminator model to ensure that the colorization has a consistent quality across all patches. The study shows that this U-Net modification can improve the generated image quality compared to a normal U-Net. The proposed architecture reaches an FID of 48.6253, SSIM of 0.8568, and PSNR of 19.7831 by only training it for 25 epochs; hence, this research offers another view of image colorization by using modules that are often used for image segmentation tasks.
Deep Learning-based Utility Pole Safety Assessment from Visual Data
Get PDFUtility poles are crucial infrastructure components, and efficiently assessing the safety of these structures and ensuring they adhere to the clearance guidelines, which specify the minimum distance between the pole and any surrounding objects, remains a challenge; the current manual inspection process is time-consuming, costly, and often subjective. This work proposes an automated deep learning-inspired solution to improve utility pole detection and measure the clearance distance. The biggest challenge was the lack of a comprehensive pole dataset; therefore, we collected a dataset containing utility poles in varied backgrounds, environments, and conditions. We compared data augmentation techniques and employed them to address the limited dataset size. The proposed approach consists of two main stages: pole detection and differentiation and pole distance measurement. The first stage is a comparison of multiple object detection models on our utility pole dataset; we used the results from the best-performing model to estimate the distance between the two pole objects. The results show that our pipeline with the YOLOv8 model outperforms SSD and achieves 83% accuracy in classifying pole compliance. The system can accurately detect and estimate clearance violations even with limited data. The success of the pipeline opens opportunities for future research; obtaining depth by using additional sensors or deep learning models could enhance the detection module. Scaling the approach to large utility pole networks while retaining real-time performance could lead to improved autonomous infrastructure maintenance
Assessing the Multifaceted Determinants of Collaborative Competence Among Students in the Digital Learning: A Comprehensive Analysis
Get PDFThe study background related to the students' collaborative competence in digital learning is still relatively low. The objective of this study is to examine the elements influencing collaborative competence. The study involved 107 cosmetology and beauty education students. The method used was a survey with data collection using questionnaires developed based on predetermined variable indicators. The analysis of data employed Structural Equation Model-Partial Least Square (SEM-PLS) with Smart PLS 4.0 software. The SEM results describe the Convergent Validity (Loading Factor and Average Variance Extracted) and Discriminant Validity (Fornell Larcker Criterion and Cross Loading), which states that the measurement model is valid. Furthermore, the Composite reliability and Cronbach's Alpha conclude that the measurement model is reliable. The analysis results indicate a positive and significant correlation of predictor variables, including project-based learning, social media, instructional approach, and material relevance to collaborative competence. Based on the variable analysis, material relevance becomes the highest aspect, followed by project-based learning, which increases the collaborative competence of students. Conversely, social media as a mediator variable weakens the level of correlation of the predictor variables to collaborative competence. This study contributes to understanding factors affecting students' collaborative competence in digital learning environments, with significant implications for educators, institutions, and policymakers in shaping digital learning frameworks enhancing collaborative competence. Future research, including longitudinal studies, could investigate the lasting impact of digital learning environments on developing collaborative competence over time
A Multi-Feature Fusion Approach for Dialect Identification using 1D CNN
Get PDFThe phonological variety of Kurdish, a language with several dialects, poses a distinct problem in automatically identifying dialects. This study examines and evaluates several sound criteria for identifying Kurdish dialects: Badini, Hawrami, and Sorani. We deployed a dataset including 6,000 samples and utilized a mix of 1D convolutional neural networks (CNN) and fully connected layers to conduct the identification job. Our study aimed to assess the efficacy of different sound characteristics in accurately identifying dialects. We employed the Mel-frequency Cepstral Coefficients (MFCC) and other features such as the Mel spectrogram, spectral contrast, and polynomial features to extract the sound characteristics. We conducted training and testing of our models utilizing both individual characteristics and a composite of all features. Our analysis revealed that the identification task achieved excellent accuracy rates, suggesting a promising potential for success. We achieved 95.75% accuracy using MFCC combined with a Mel spectrogram. The accuracy improved by including contrast in the MFCC feature extraction process to 91.42%. Similarly, using poly_features resulted in an accuracy of 90.83%. Remarkably, accuracy reached a maximum of 96.5% when all the attributes were combined
Uncovering the Most Effective Pedagogical Techniques for Math Education Using Machine Learning
Get PDFMany new math educators express that their first years in the teaching field are extremely challenging. They struggle to discover and apply the most effective teaching techniques and behaviors, often without the support of more experienced colleagues. In this study, we use machine learning to find the strategies that novice teachers can adopt to enhance their teaching effectiveness. The core aim is to uncover the relationship between teachers’ performance, as assessed by student evaluations, and their pedagogical methods. These strategies are derived from the final decision tree model, which is trained on a large dataset of empirical data from schools. The data consists of input from 72 math teachers of grades 7-9 and their students in Dubai, used to train two decision tree models: a classification tree and a regression tree. The structure of these trees is analyzed to identify and rank the effectiveness of nine teaching techniques—such as Visualization, Practice, Math Rules, Gamification, Collaboration, Problem-Solving, Case Studies, Assessments, and Language Switching—and four behavioral methods—such as Inspiration, Engagement, Entertainment, and Bonding— in relation to the Student Evaluation Index (SEI), which is derived from student feedback. Results indicate that techniques such as "gamification" and "inspirational behavior" are consistently associated with higher SEI scores across different tree configurations. However, factors such as the demographics and culture of both students and teachers may need to be considered when generalizing these findings to other regions of the world
Development of Augmented Reality Based Interactive Learning Media on Electric Motor Installation Subjects
Get PDFInteractive learning media play an essential role in the learning process. In today's digital era, interactive learning media is one of the best choices for improving the quality of learning by making it more effective and efficient. This study aims to develop augmented reality (AR)--based interactive learning media for electric motor installation subjects. The method used in this research is the research and development (R&D) method with the 4-D development model. This research consists of several stages: defining, designing, developing, and disseminating. Based on the research results, it can be concluded that this augmented reality learning media development research has produced a valid and practical augmented reality learning medium for electric motor installation subjects. The media and material validation results consist of two media validators and two material validators. Obtained a media validation value of 0.58 and a material validation value of 0.62 using the validity interval category ≥ 0.4. Thus, the media and materials developed were declared valid. The augmented reality learning media practicality test results obtained 86.63% by using the practicality interval category > 75%–100%. Thus, AR-based learning media in electric motor installation have been considered practical. Implications for further research could include developing AR media for other subjects or applying AR technology in various learning contexts, including integrating more complex interactive features or adapting the media for broader educational needs. In addition, this research could encourage exploring AR in distance learning scenarios and increase student engagement through gamification or further simulation in skills-based learning
Drone Kit-Python for Autonomous Quadcopter Navigation
Get PDFUsing Python scripts over the MAVLink protocol, developers can use the open-source DroneKit Python software framework to enable autonomous drone operations. This framework provides excellent flexibility and power to facilitate automated drone control. The built quadcopter has an X configuration and uses a DJI F450 frame with some modifications. Interestingly, the drone has legs made of aluminum on both sides to help with smooth takeoffs and landings. The frame is 45 cm diagonal length and 30 cm vertical height. The drone was given an additional weight in a 15 x 18 x 12.5 cm box. The propeller used in this investigation is a 9x6 carbon-based model. The x2216 1400kV brushless motor that is being used is from Sunnysky, and it comes with an Electronic Speed Controller (ESC) with a 30A rating. A 4-cell 14.8V Lithium-Polymer (Li-Po) battery with a 7200mAh capacity powers the drone. Apart from that, the drone weighs 1573g in total. The results are obtained by self-measurement and flight measurement data (FMU). Six attempts were made, and the results showed that the second flight had the longest flight time and the highest altitude. In particular, the Flight Measurement Unit (FMU) reported that the flight lasted 81 seconds and reached an altitude of 0.93 meters. In contrast, the self-measurement data reported that the flight lasted 85 seconds and reached an altitude of 1.5 meters
Visual Analytic for Traffic Impact Assessment
Get PDFThis study strives to promote the state of traffic impact assessment through high-end visual analytics by incorporating spatial and temporal data visualization to enhance traffic management. Based on a dataset on traffic flow at three major intersections, we married data cleaning, integration, and transformation to set out for a detailed visual analysis. Thus, the critical materials comprise the traffic count in multiple lanes, vehicle types, and saturation flow rates to understand the road network's capacity. They essentially explored the traffic volume variations daily and hourly and pattern identification using heat maps, parallel coordinate charts, and bar plots. Thus, the findings expose the remarkable traffic volume and pattern differences by distinguishing peak and off-peak hours on weekdays and weekends. The level of service at each junction was determined by the volume-to-capacity ratio, identifying potential congested areas. As such, this work points to the importance of further improvements to visual analytic techniques to accurately predict traffic patterns and evaluate traffic management strategies effectively. Predictive models based on visual analytic findings can pave the way for proactive traffic control and congestion mitigation, making urban traffic management more efficient and safer. The current study provides a scaffold for additional exploration of the above-detailed methods and their penal outcomes in urban development planning and policy provision in terms of developing sustainable traffic control strategies and real-time decision-making improvements
Design of Tools for Visualizing Thermodynamic Concepts in Steam Power Plant Trainer Processes with Web-Based Exploratory Data Analysis (EDA)
Get PDFThermodynamics is considered one of the most complex and challenging subjects for many students. This is primarily due to comprehending abstract concepts such as entropy, enthalpy, and energy flow, which involve complex mathematical equations and are rarely accompanied by tangible visualizations. This research aims to design, develop, and test a data-based visualization tool for thermodynamics testing results. This study collected and processed data from thermodynamics testing and simulations, such as the mini-steam power plant trainer used as a teaching aid in thermodynamics education, as the foundation for designing a data-based visualization tool for thermodynamics concepts. The visualization tool was created using the Python programming language integrated with the web-based Streamlit framework. The designed visualization tool encompasses various features, including automated data reporting, visualization of variable correlations using correlation heatmaps, Sankey diagrams for visualizing energy flow, and the capability to predict electrical output using machine learning integrated with three different machine learning algorithms. The visualization tool was evaluated by thermodynamics experts using a Likert scale. Based on the results obtained, the experts gave an average score of 4 in the information accuracy aspect in the good category. This shows that the information displayed in this visualization tool is by thermodynamics learning at Padang State University. In the visualization aspect, experts gave an average score of 4.25, which is in the Good and Very Good range. In alignment with the education aspect, experts gave an average score of 3.75, which is close to the good category. This shows that this aspect is considered suitable for studying thermodynamics, although shortcomings still need to be corrected. Experts gave a relatively high assessment of the Ease-of-Use aspect, with an average score of 4.5, with a range of Good and Very Good. This enables students to better understand complex patterns, cause-and-effect relationships, and parameter changes within thermodynamics concepts
Dynamic Key Generation Using GWO for IoT System
Get PDFOne well-known technological advancement that significantly impacts many things is the Internet of Things (IoT). These include connectivity, work, healthcare, and the economy. IoT can improve life in many situations, including classrooms and smart cities, through work automation, increased output, and decreased worry. However, cyberattacks and other risks significantly impact intelligent Internet of Things applications. Key generation is essential in information security and the various applications that use a distributed system, networks, or Internet of Things (IoT) systems. Several algorithms have been developed to protect IoT applications from malicious attacks; since IoT devices usually have small memory resources and limited computing and power resources, traditional key generation methods are inappropriate because they require high computational power and memory usage. This paper proposes a method of Dynamic Key Generation Method (DKGM) to overcome the difficulty using a specific chaotic map called the Zaslavskii Map and a swarm intelligent algorithm for optimization called Grey Wolf Optimizer (GWO). DKGM's ability to generate several groups-seed numbers using the Zaslavskii map depends on various initial parameters. GWO selects strong generated numbers depending on the randomness test as a fitness function. Three wolfs GWα, GWβ, and GWΩ, are used to simulate the behavior of a pack of grey wolves when attacking prey. The speed and position of each wolf are updated depending on the best three wolves. Finally, use the sets GWα in the round, GWβ in the subkey, and GWΩ in shifting operations of the Chacha20 hash function. The dynamic procedure was used to improve the high-security analysis of the DKGM approach over earlier methods. Simulations show that the suggested method is preferable for IoT applications