International Journal of Informatics and Communication Technology (IJ-ICT)
Not a member yet
494 research outputs found
Sort by
IndoBART optimization for question answer generation system with longformer attention
Get PDFThe Incorporation of Question Answering system holds immense potential for addressing Indonesia’s educational disparities between the abundance of high school students and the limited number of teachers in Indonesia. These studies aim to enhance the Question Answering System model tailored for the Indonesian language dataset through enhancements to the Indonesian IndoBART model. Improvement was done by incorporating Longformer’s sliding windows attention mechanism into the IndoBART model, it would increase model proficiency in managing extended sequence tasks such as question answering. The dataset used in this research was TyDiQA multilingual dataset and translated the SQuADv2 dataset. The evaluation indicates that the Longformer-IndoBART model outperforms its predecessor on the TyDiQA dataset, showcasing an average 26% enhancement across F1, Exact Match, BLEU, and ROUGE metrics. Nevertheless, it experienced a minor setback on the SQuAD v2 dataset, leading to an average decrease of 0.6% across all metrics
Assessing the user experience of marker-based 3D WebAR applications using user experience questionnaire
Get PDFMarker-based 3D web-based augmented reality (WebAR) applications are an emerging field that merges web technologies with augmented reality. WebAR has gained popularity because of its ability to provide users with a reliable and autonomous platform. Yet, a limited investigation has verified its application and user perspective on its ability to function. This study is designed to evaluate the user experiences of marker-based 3D WebAR applications using the user experience questionnaire (UEQ). This study assesses various elements of the user experience, including attractiveness, clarity, engagement, efficiency, and innovation, utilizing the UEQ. This study aims to analyze user perceptions and interaction patterns thoroughly to get useful insights into the usability and user satisfaction aspects of marker-based 3D WebAR apps. The findings reveal that the WebAR app is both appealing and efficient, instilling confidence in its users. This underscores the pivotal role of user experience in shaping the effectiveness and reception of WebAR applications. This research has the potential to influence the creation of more user-focused and engaging marker-based 3D WebAR experiences, improving user engagement and immersion in web-based augmented reality environments
Strategic Deployment of EV Charging Infrastructure: An In-Depth Exploration of Optimal Location Selection and CC-CV Charging Strategies
Get PDFThe continued expansion of the electric vehicle (EV) market necessitates strategic planning for the placement of charging stations to ensure efficient access and utilization of electric infrastructure. This paper presents a comprehensive review of the critical factors in optimizing the selection of EV charging station locations, along with the implementation of Constant Current-Constant Voltage (CC-CV) charging models. The study addresses the challenges and opportunities in identifying the most effective locations for charging stations to accommodate the growing demand for sustainable transportation. Furthermore, it examines the benefits of adopting CC-CV charging models to improve the charging process, achieving a balance between charging speed and battery longevity. Through this analysis, the review aims to provide valuable insights to stakeholders involved in the development and expansion of EV charging infrastructure, thereby supporting the transition to a more sustainable and extensive electric mobility ecosystem
Comparative study of traditional edge detection methods and phase congruency based method
Get PDFFinding relevant and crucial details from images and effectively interpreting what they represent are two of image processing's main goals. An edge is the line that separates an object from its backdrop and shows where two things meet. Mining the picture's borders for extracting useful data remains one of the trickiest steps in understanding of an image. The borders of the objects may be used to build the image's edges, which are its basic characteristics. There are different types of traditional edge retrieval techniques that are conventionally categorized as first order and second gradient based methods such as Roberts, Prwitt, Kirsch, Robinson, canny, Laplacian and Laplacian of gaussian. The majority of research and review work on edge detection algorithms focuses on conventional algorithms and soft computing based methods, neglecting illumination invariant phase congruency based edge detector. This study aims to compare traditional derivative based edge detection algorithms with log Gabor wavelet based edge detector phase congruency. This work does a thorough examination of various edgedetecting approaches, including traditional boundary detection methods and log Gabor wavelet based method. To test effectiveness of edge detection algorithms, experimental results are obtained on images from DRIVE, STARE, and BSDS500 dataset
Soil moisture prototype soil moisture sensor YL-69 for Gaharu (Aquilaria malaccensis) tree planting media
Get PDFSoil moisture, defined as the amount of water present in the spaces between soil particles, plays a critical role in plant growth. Excessive soil moisture can lead to issues such as root rot, deviating from the ideal conditions required for root absorption. To address this, we developed a prototype tool using the YL-69 soil moisture sensor to monitor and control the soil moisture levels in Agarwood/Gaharu tree planting media. The prototype was designed to activate a water pump when soil moisture exceeded 80%, ensuring optimal humidity for plant growth. Once the moisture level dropped below 80%, the pump was deactivated to prevent overwatering. The YL-69 sensor demonstrated an accuracy of 88.76% under controlled conditions. This study highlights the potential of using low-cost sensors for automated soil moisture management in small-scale Gaharu cultivation
Comparative analysis of u-net architectures and variants for hand gesture segmentation in parkinson’s patients
Get PDFU-Net is a well-known method for image segmentation, and has proven effective for a variety of segmentation challenges. A deep learning architecture for segmenting hand gestures in parkinson’s disease is explored in this paper. We prepared and compared four custom models: a simple U-Net, a three-layer U-Net, an auto encoder-decoder architecture, and a U-Net with dense skip pathways, using a custom dataset of 1,000 hand gesture images and their corresponding masks. Our primary goal was to achieve accurate segmentation of parkinsonian hand gestures, which is crucial for automated diagnosis and monitoring in healthcare. Using metrics including accuracy, precision, recall, intersection over union (IoU), and dice score, we demonstrated that our architectures were effective in delineating hand gestures under different conditions. We also compared the performance of our custom models against pretrained deep learning architectures such as ResNet and VGGNet. Our findings indicate that the custom models effectively address the segmentation task, showcasing promising potential for practical applications in medical diagnostics and healthcare. This work highlights the versatility of our architectures in tackling the unique segmentation challenges associated with parkinson’s disease research and clinical practice
Empowering low-resource languages: a machine learning approach to Tamil sentiment classification
Get PDFSentiment analysis is essential for deciphering public opinion, guiding decisions, and refining marketing strategies. It plays a crucial role in monitoring public sentiment, fostering customer engagement, and enhancing relationships with businesses' target audiences by analyzing emotional tones and attitudes in vast textual data. Sentiment analysis is extremely limited, particularly for languages like Tamil, due to limited application in diverse linguistic contexts with fewer resources. Given its global impact and linguistic diversity, addressing this gap is crucial for a more nuanced understanding of sentiments in India. In the context of Tamil, the need for sentiment analysis models is particularly crucial due to its status as one of the classical languages spoken by millions. The cultural, social, and historical nuances embedded in Tamil language usage require tailored sentiment analysis approaches that can capture the subtleties of sentiment expression. This paper introduces a novel method that assesses the performance of various text embedding methods in conjunction with a range of machine learning (ML) algorithms to enhance sentiment classification for Tamil text, with a specific focus on lyrics. Experiments notably emphasize FastText word embedding as the most effective method, showcasing superior results with a remarkable 78% accuracy when coupled with the support vector classification (SVC) model
Deep learning for grape leaf disease detection
Get PDFAgriculture is crucial to India's economy. Agriculture supports almost 75% of the world's population and much of its gross domestic product (GDP). Climate and environmental changes pose a threat to agriculture. India is recognized for its grapes, a commercially important fruit. Diseases reduce grape yields by 10-30%. If not recognized and treated early, grape diseases can cost farmers a lot. The main grape diseases include downy and powdery mildew, leaf blight, esca, and black rot. This work creates an Android grape disease detection app which uses machine learning. When a farmer submits a snapshot of a diseased grape leaf, the smartphone app identifies the ailment and offers grape plant disease prevention tips. In this research, an android app that detects grape plant illnesses use convolutional neural network (CNN) and AlexNet machine learning architectures. We investigated and compared CNN and AlexNet architecture's efficacy for grape disease detection using accuracy and other metrics. The dataset used comes from Kaggle. CNN and AlexNet architectures yielded 98.04% and 99.03% accuracy. AlexNet was more accurate than CNN in the final result
Bolstering image encryption techniques with blockchain technology - a systematic review
Get PDFMultimedia data plays a momentous role in present world. With the advancements in various fields of research like internet of things (IoT), industrial IoT (IIoT), cloud computing, medical image processing, and many more technologies, the digital images have already encroached the multimedia eon. The major challenge lies in providing a tamper proof image with higher level of security and confidentiality while being transmitted through a public network. Image encryption techniques are considered to be the predominant method to anticipate security from any unauthorized user access. This has indeed provoked the researchers to create new diverse and hybrid algorithms for encrypting the images. At present blockchain has been the most prevalently discussed method for security and the next level of security can be foreseen using the blockchain encryption techniques. This paper identifies the literature which mainly focuses on assorted image encryption techniques with blockchain technology applied on digital images from heterogeneous sources. An overview has been proposed to discuss on these techniques
Prediction and classification of diabetic retinopathy using machine learning techniques
Get PDFDiabetic retinopathy (DR) is a progressive and sight-threatening complication of diabetes mellitus, characterized by damage to the blood vessels in the retina. Early detection of DR is vital for timely intervention and effective management to prevent irreversible vision loss. This paper provides a comprehensive review of recent advancements in integrating machine learning (ML) and deep learning (DL) techniques for diagnosing DR, aiming to assist ophthalmologists in their manual diagnostic process. The paper presents a comprehensive definition of DR, elucidating the underlying pathological processes, clinical signs, and the various stages of DR classification, ranging from mild non-proliferative to severe proliferative DR. Integrating ML and DL in DR diagnosis has developed the field by offering automated and efficient methods and techniques to analyze retinal images. With high sensitivity and specificity, these techniques demonstrate their efficacy in accurately identifying DR-related lesions, such as microaneurysms, exudates, and hemorrhages. Furthermore, the paper examines diverse datasets employed in training and evaluating ML and DL models for DR diagnosis. These datasets range from publicly available repositories to specialized datasets curated by medical institutions. The role of large-scale and diverse datasets in enhancing model robustness and generalizability is emphasized