Електронний архів-репозитарій
Одеського національного університету імені І. І. Мечникова
ISSN:2310-7731
Вітаємо на цифровій платформі elONUar, що забезпечує накопичення, систематизацію, обробку, зберігання та надання у відкритий доступ електронних версій наукових, науково-дослідних, навчально-методичних праць та кваліфікаційних робіт наукових та науково-педагогічних співробітників, аспірантів та студентів Одеського національного університету імені І. І. Мечникова, а також електронних версій університетських друкованих видань.
Супровід та підтримка здійснюються Науковою бібліотекою університету
Нові матеріали для розміщення надсилайте на e-mail dspace@onu.edu.ua
Фонди
Виберіть фонд, щоб переглянути його зібрання.
Нові надходження
Information Technology for Optimization of High-Resolution Remote Sensing Image Semantic Segmentation Based on Self-Attention Mechanism
(Одеський національний університет імені І. І. Мечникова, 2024) Юй Фей; Yu Fei
This study explores detail optimization for semantic segmentation of remote sensing images with high-resolution, emphasizing the use of self-attention mechanism to alleviate detail loss and improve segmentation accuracy and robustness. of segmentations. In this paper, the principle of the self-attention mechanism is explored, an innovative semantic segmentation model is designed and implemented, and its effectiveness is verified by experiments, all of which could provide theoretical guidance and practical support for technological development.
In the experimental section, the proposed method was applied to several datasets: PASCAL VOC, Cityscapes, and COCO. The results showed that the selfattention mechanism based improved model had better performance in image segmentation tasks. Especially, on the Cityscapes dataset, the mAP reached 90.2%, which attests to the effectiveness of the approach in handling urban scenes. Besides, an evaluation of various loss functions revealed that hybrid loss functions consistently outperformed others in enhancing object detection performance. These achievements not only enhanced the precision of semantic segmentation for remote sensing images with high-resolution but also strengthened the model's adaptability to complex scenarios.
Information technology of the Optimizing Internal Railway Transportation Paths in Metallurgical Enterprises Using Dijkstra's Algorithm
(Одеський національний університет імені І. І. Мечникова, 2024) Pengyang Liu
This paper explores the theme of "Information Technology Based on Dijkstra Algorithm in Metallurgical Railways." It discusses the application of the Dijkstra algorithm to optimize the selection of railway transportation paths in the metallurgical industry, aiming to enhance transportation efficiency, reduce costs, and alleviate the workload of internal railway employees in metallurgical enterprises. With the continuous advancement of the metallurgical industry, the growth in production volume has led to a significant increase in the frequency of molten iron transportation, making the role of railway transportation increasingly important. However, the complex layout of railway lines within metallurgical plants, numerous switches, coupled with the high speed of shunting and the diversity of transportation demands, pose considerable challenges for railway path planning.
The Dijkstra algorithm, as an efficient tool for shortest path searches, has demonstrated significant effectiveness in addressing route planning issues. Its potency stems from its ability to manage single source shortest path problems, its path recording mechanism, and its graphical representation advantages. The essence of this algorithm lies in its gradual expansion process: starting from the origin, it visits nodes in the graph one by one and calculates the shortest path from the starting point to each node. The Dijkstra algorithm not only effectively identifies the shortest path but also records the specific details of the path, providing comprehensive reference information for subsequent path selection.
This paper initially outlines the core concepts of the Dijkstra algorithm and its applicability in the field of route planning. Following that, a path optimization strategy based on the Dijkstra algorithm is designed, tailored to the specific characteristics of metallurgical railway transportation. This strategy takes into account not only the distance aspect of the path but also comprehensively considers multiple dimensions, including safety, economic cost, time efficiency, and expansion potential, all in an effort to find the optimal transportation solution.
By constructing a mathematical model and combining it with specific practical cases, the algorithm's application value and practical effectiveness are verified.
In practical applications, the Dijkstra algorithm can effectively tackle various issues encountered in metallurgical railway transportation. For instance, it can assist dispatchers in selecting the optimal path by accurately calculating the cost of each path, thereby reducing unnecessary travel distances and saving time and resources. Moreover, this method exhibits excellent flexibility and adaptability, capable of adjusting relevant parameters according to different transportation requirements, providing an efficient and stable solution for railway transportation in the metallurgical industry. For example, when rapid response to emergencies or adjustments to transportation plans are needed, the Dijkstra algorithm can quickly recalculate the path to ensure the smooth progress of transportation tasks.
Furthermore, the application of the Dijkstra algorithm also lays the groundwork for the construction of subsequent intelligent dispatching systems. By integrating with big data analysis, Internet of Things (IoT) technology, and Artificial Intelligence (AI), the Dijkstra algorithm can achieve dynamic path optimization in more complex environments. For example, using real time data monitoring and predictive analysis, potential transportation bottlenecks can be identified in advance, allowing for preventive measures; through machine learning algorithms, path selection strategies can be automatically adjusted, further enhancing the intelligence and precision of path planning.
In summary, path optimization using the Dijkstra algorithm not only enhances the precision and efficiency of path planning but also provides more flexible and reliable technical support for railway transportation in the metallurgical industry. This contributes to enhancing the competitiveness of enterprises and promoting the sustainable development of the entire industry. In the future, with the continuous advancement of information technology, the Dijkstra algorithm will play a greater role in metallurgical railway transportation, promoting the diversified development of shortest path algorithm selection.
Інтелектуальна стратегія оптимізації обслуговування клієнтів з урахуванням системи ОПМ діалогу
(Одеський національний університет імені І. І. Мечникова, 2024) Пан Гуанхуй; Pang Guanghui
With the continuous expansion of digital services, companies are increasingly relying on intelligent customer service systems to improve user experience and efficiency. However, traditional dialogue systems often fail to meet user expectations due to insufficient comprehension, response delays, and low personalisation. The purpose of this research is to analyze the limitations of the existing intelligent customer service system and propose specific optimization strategies to improve the natural language understanding (NLV), dialogue management (PM), and response generation (DG) capabilities of the dialogue system. By using the latest artificial intelligence technology and algorithms, the research results showed that the optimization strategy significantly improved the system's response speed, accuracy, and user satisfaction. This paper not only provides theoretical support for the development of intelligent customer service system, but also provides strategic guidance for practical operation.
Information Technology for Optimization of Common Cold Weapon Object Detection Algorithm Based on YOLOv11
(Одеський національний університет імені І. І. Мечникова, 2024) Li Yongming
Under the theme of "Optimization of the cold weapon detection algorithm based on YOLOv11", this paper explores the pressing demand for enhancing the accuracy and efficiency of cold weapon detection against the backdrop of increasingly severe public security issues. This study is particularly important in view of the increasing number of criminal incidents involving cold weapons, which pose a major threat to social stability and personal safety.
The study began by collecting and managing a comprehensive dataset of cold weapons, leveraging the e-commerce platform's vast repository to ensure a diverse and representative collection of weapon types. This dataset contains a variety of cold weapons and is the basis for training and evaluation of the
YOLOv11 algorithm. The dataset is carefully curated to reflect the variability of cold weapon appearance, including different models, conditions, and backgrounds, thus providing a solid foundation for algorithm training and testing.
Through several strategy adjustments, the optimization of YOLOv11 algorithm is realized. The configuration of the network layer is finely adjusted to boost the model's capability of extracting cold weapon features. Advanced loss functions are utilized to improve the model's capacity to generalize from training data and heighten its robustness against changes in real-world scenarios. These modifications were critical to adapting the YOLOv11 algorithm to the nuances of cold weapon detection, resulting in significant improvements in its detection capabilities.
A key contribution of this study is the introduction of focal - iou loss function, which solves the class imbalance problem prevalent in target detection datasets. This innovative loss function not only provides a more comprehensive assessment of the quality of the prediction box, but also adjusts the loss weights to focus the model's attention on samples that are harder to classify. The integration of the GSConv module represents another important advance, as it simplifies the calculation process without compromising the accuracy of the model. This module offers a more efficient feature extraction approach via grouping and separable volume design, thereby making the algorithm more applicable to real-time applications.
The results of the experiments carried out in this study indicate that the optimized YOLOv11 algorithm (called YOLOv11s+) has superior performance in various indexes. Compared to the previous model and other models, YOLOv11s+ has higher accuracy, recall and average accuracy (mAP) at different cross over association (IoU) thresholds. These metrics play a critical role in assessing the efficacy of target detection algorithms, particularly in situations where detection accuracy is of vital significance.
In a word, the research not only promotes the technological frontier of cold weapon detection, but also has practical significance for strengthening public safety. The optimized YOLOv11 algorithm provides a promising solution for accurate and efficient detection of cold weapons, and has potential application prospects in smart city development and social governance.
Information Technology of Learning Personalization based on the Advanced Big Data Analytics Methods
(Одеський національний університет імені І. І. Мечникова, 2024) Han Zaihui
The thesis explores the transformative potential of information technology of learning personalization based on the advanced Big Data analytics methods. It underscores the significance of leveraging advanced BDA techniques to enhance educational outcomes by tailoring learning experiences to individual needs. The study evaluates various BDA tools and methodologies, assessing their efficacy in processing and analyzing vast educational datasets. The research employs a mixed-method approach, integrating quantitative data analysis with qualitative case studies. The findings suggest that BDA can significantly improve personalized learning, leading to better academic performance and student satisfaction. The thesis concludes with recommendations for educational institutions to integrate BDA into their learning management systems to facilitate personalized learning pathways.
In the wake of the digital transformation catalyzed by the COVID-19 pandemic, educational institutions have witnessed an unprecedented surge in the adoption of online learning platforms. This shift has resulted in an explosion of educational data, providing a rich repository of information that can be harnessed to enhance the learning experience. Big Data Analytics (BDA) plays a pivotal role in this context, offering insights into student behaviors, preferences, and performance patterns that can inform the design of personalized learning pathways.
The purpose of this thesis is to investigate the role of advanced BDA in revolutionizing personalized learning and its impact on educational outcomes post-2020. The study aims to understand how BDA can be utilized to process and analyze educational data to create tailored learning experiences that address the unique needs of individual students. The research focuses on the evaluation of various BDA tools and methodologies, including Machine Learning (ML), Artificial Intelligence (AI), and Educational Data Mining (EDM), to assess their effectiveness in enhancing personalized learning.
A mixed-method research approach is adopted, combining quantitative data analysis with qualitative case studies to provide a comprehensive understanding of 3
the impact of BDA on personalized learning. The quantitative analysis involves the examination of large educational datasets from various learning management systems (LMS), while the qualitative component includes detailed case studies of educational institutions that have successfully integrated BDA into their teaching and learning strategies.
The case studies reveal that the integration of BDA in educational practices has led to significant improvements in student academic performance and satisfaction. BDA enables educators to identify at-risk students early, personalize content delivery, and adapt teaching methods to better meet the needs of diverse learners. Additionally, BDA facilitates the creation of dynamic learning environments that can adapt in real-time to student interactions and feedback, thus enhancing the overall learning experience.
The thesis also discusses the challenges associated with the implementation of BDA in education. These include issues related to data privacy, the need for robust data infrastructure, and the requirement for technical expertise to manage and analyze the data effectively. Despite these challenges, the opportunities presented by BDA are substantial, with the potential to revolutionize education by making it more accessible, inclusive, and effective.
Природна характеристика гирлової області Дунаю
(Одеський національний університет імені І. І. Мечникова, 2024) Окунєв, Михайло Дмитрович
Найголовнішими водними обєктами України є річки. Тут протікают одні із найбільших річок Європи – Дніпроі Дунай. Всі річки Укрпїни належать до басейну Атлантичного океану і більшість із них несуть свої води до безприпливного Чорного морч. При впадінні в море вони формують різні за походженням гирлові області, але більшість із них це лимани. Дунай сформував одну із найбільших в Європі дельту, яка пройшла декілька етапів в своєму розвитку. В наш час Дунай формує Кілійську дельту висунення, яка є по декільким показникам унікальним природним творінням. Дельта є місцем гніздування мігруючих птиць, в водах проток Кілійської дельти водяться рідкісні риби, острова і водно-болотнІ угіддяя – притулок ендемічних вилів рослин і тварин.
Оцінка земель сільськогосподарського призначення в умовах воєнного стану (на прикладі Первомайського району Миколаївської області)
(Одеський національний університет імені І. І. Мечникова, 2024) Маркасов, Дмитро Михайлович
Земля є невід’ємною частиною людського існування. У відповідності до законодавства України «земля, її надра … є об'єктами права власності Українського народу... земля є основним національним багатством, що перебуває під особливою охороною держави…» [14]. Як основний базис усіх процесів життєдіяльності суспільства земля виконує ряд функцій, зокрема як природний об’єкт, що охороняється законом, існує незалежно від волі людини, виконує екологічну функцію; як місце й умова життя людини – соціальну; як територія держави – політичну; як об’єкт господарювання – економічну функції; крім того земля має вартість, оцінка якої є необхідною умовою нормального функціонування і розвитку економіки будь-якої країни.
Analysis and Implementation of Objiect Detection Information Technology Based on Deep Learning
(Одеський національний університет імені І. І. Мечникова, 2024) Фен Сюецзяо; Feng Xuejiao
The topic of "Analysis and Implementation of Object Detection Information Technology Based on Deep Learning" is considered in the thesis. Deep learning-based object detection technologies have become widespread, particularly in the field of autonomous driving and intelligent transportation systems. These technologies offer significant advantages in terms of accuracy and efficiency, enabling real-time detection and recognition of various objects, including vehicles, pedestrians, and traffic signs. However, despite their advantages, such systems face challenges in complex environments, such as foggy weather, low light conditions, and dense traffic scenarios, which can degrade detection performance.
The purpose of the work is to improve existing deep learning-based object detection algorithms to enhance their robustness and accuracy in challenging conditions.
As a result of the research carried out in the work, various state-of-the-art object detection algorithms were analyzed, and novel methods for improving detection performance in foggy conditions and lightweight network architectures were proposed. Specifically, the integration of dehazing techniques, attention mechanisms, and optimized loss functions were explored to enhance the detection accuracy and efficiency of the YOLOv5 and YOLOv8 models. Additionally, the work included the construction of a foggy scene vehicle detection dataset and the implementation of a model system for simulating and evaluating the proposed methods.
In the work, a number of requirements for the class library, which can be used in deep learning-based object detection systems, as well as for the model system, which made it possible to simulate the considered methods, were formed. Both systems were implemented.
Using the proposed methods, it was possible to achieve significant improvements in object detection accuracy under foggy conditions and in lightweight network architectures. The integration of dehazing techniques and attention mechanisms allowed for better feature extraction and focus on key areas, while the optimized loss functions improved the model's ability to detect small and occluded objects. The lightweight network architectures demonstrated a good balance between computational efficiency and detection accuracy, making them suitable for deployment on resource-constrained devices.
Overall, the research provides valuable insights and practical solutions for enhancing the performance of deep learning-based object detection systems in realworld applications.
Грунтово-земельні ресурси Коноплянської сільської територіальної громади Березівського району Одеської області та їх використання
(Одеський національний університет імені І. І. Мечникова, 2024) Малюта, Юрій Миколайович
Використання ґрунтів і земельних ресурсів нині є досить актуальним питанням. Часто наразі можна зустріти термін «ґрунтово-земельні ресруси» (ҐЗР), що характеризує ґрунти, як важливу складову біосфери (педосферу), забезпечуючи життєдіяльність екосистем та сталого розвитку людства. Вони є основним засобом сільськогосподарського і лісогосподарського виробництва, однак постійно знаходяться під впливом антропогенного тиску, не завжди контрольованого, регламентованого та зваженого. Управління сталим викристанням потенціалу ҐЗР особливо важливо в умовах сучасних викликів, пов’язаних з глобальними та локальними змінами клімату, впливом війни, складністю дотримання сталих агропрактик в умовах невизначеності, тощо.
Методи та інформаційні технології масштабування web додатків
(Одеський національний університет імені І. І. Мечникова, 2024) Камєнєв, Кирило Ігорович
У цьому дослідженні розглядаються сучасні методи та технології масштабування веб-додатків, що дозволяють вирішувати проблеми, пов'язані зі зростаючим попитом та мінливим трафіком. У дослідженні проаналізовано підходи до вертикального та горизонтального масштабування, використання розподілених баз даних, мікросервісів та безсерверних архітектур, з акцентом на хмарні платформи, такі як AWS Lambda та Amazon Aurora. Особливу увагу приділено практичному застосуванню та оцінці безсерверних рішень для досягнення ефективної масштабованості. Висновки надають структуровану оцінку стратегій масштабування, акцентуючи увагу на архітектурних компромісах та їхньому впливі на продуктивність і оптимізацію ресурсів у веб-додатках.