Joint Diagnosis of Pneumonia, COVID-19, and Tuberculosis from Chest X-ray Images: A Deep Learning Approach

Diagnostics, Journal Year: 2023, Volume and Issue: 13(15), P. 2562 - 2562

Published: Aug. 1, 2023

Pneumonia, COVID-19, and tuberculosis are some of the most fatal common lung diseases in current era. Several approaches have been proposed literature for diagnosis individual diseases, since each requires a different feature set altogether, but few studies joint diagnosis. A patient being diagnosed with one disease as negative may be suffering from other disease, vice versa. However, said related to lungs, there might likelihood more than present same patient. In this study, deep learning model that is able detect mentioned chest X-ray images patients proposed. To evaluate performance model, multiple public datasets obtained Kaggle. Consequently, achieved 98.72% accuracy all classes general recall score 99.66% 99.35% No-findings, 98.10% Tuberculosis, 96.27% respectively. Furthermore, was tested using unseen data augmented dataset proven better state-of-the-art terms metrics.

Language: Английский

---

Predictive analysis-based sustainable waste management in smart cities using IoT edge computing and blockchain technology

Computers in Industry, Journal Year: 2025, Volume and Issue: 166, P. 104234 - 104234

Published: Jan. 5, 2025

Language: Английский

---

Personal Protective Equipment Detection: A Deep-Learning-Based Sustainable Approach

Sustainability, Journal Year: 2023, Volume and Issue: 15(18), P. 13990 - 13990

Published: Sept. 20, 2023

Personal protective equipment (PPE) can increase the safety of worker for sure by reducing probability and severity injury or fatal incidents at construction, chemical, hazardous sites. PPE is widely required to offer a satisfiable level not only protection against accidents aforementioned sites but also chemical hazards. However, several reasons negligence, workers may commit comply with regulations wearing equipment, occasionally. Since manual monitoring laborious erroneous, situation demands development intelligent systems automated real-time accurate detection compliance. As solution, in this study, Deep Learning Computer Vision are investigated near detection. The four colored hardhats, vest, glass (CHVG) dataset was utilized train evaluate performance proposed model. It noteworthy that solution detect eight variate classes PPE, namely red, blue, white, yellow helmets, head, person, glass. A two-stage detector based on Fast-Region-based Convolutional Neural Network (RCNN) trained 1699 annotated images. model accomplished an acceptable mean average precision (mAP) 96% contrast state-of-the-art studies literature. study potential contribution towards avoidance prevention fatal/non-fatal industrial means real-time.

Language: Английский

---

E-WASTE MANAGEMENT THROUGH DEEP LEARNING: A SEQUENTIAL NEURAL NETWORK APPROACH

FUDMA Journal of Sciences, Journal Year: 2024, Volume and Issue: 8(3), P. 17 - 24

Published: July 29, 2024

The goal of this research is to improve the management electronic trash (e-waste) by using a Sequential Neural Network (SNN) with TensorFlow and Keras as part an advanced deep learning technique. In order address growing problem e-waste, collects large amount data from images e-waste then carefully preprocesses augments those images. With precision, recall, F1 scores 87%, 86%, respectively, SNN architecture—which incorporates dropout, pooling, convolutional layers—achieved amazing 100% classification accuracy. These outstanding outcomes show how well model can classify components, suggesting that it has potential be used in real-world scenarios. results indicate SNN-based approach greatly improves accuracy efficiency sorting, promoting environmental sustainability resource conservation. By automating sorting process, suggested system decreases need for manual labor, minimizes human error, speeds up processing. study emphasizes model's suitability integration into current workflows, providing scalable dependable way expedite recycling process. Additionally, real-time applicability highlights its revolutionize practices, making positive ecological impact. . Future endeavors will center on broadening dataset include wider range image categories, investigating more architectures, incorporating Internet Things (IoT) devices monitoring management.

Language: Английский

---

SwinConvNeXt: a fused deep learning architecture for Real-time garbage image classification

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 7, 2025

Abstract Waste management handles all kinds of waste, including household, industrial, municipal, organic, biomedical, biological, and radioactive wastes. People still face challenges in proper disposal methods for different types landfill-bound items, recyclable materials, biodegradable waste. Inadequate waste poses a significant multifaceted global challenge. The conventional method segregating is time-consuming ineffective that wastes human power money. To address this issue real time, sophisticated sustainable systems need to be implemented. latest advancements computer vision deep learning offer efficient solutions effective recycling management. Existing models exhibited various limitations, such as detection accuracy computational inefficiency, particularly when dealing with objects varying sizes exhibiting high degrees visual similarity. These limitations generate effectively capturing representing the nuanced features visually similar objects. problem, we proposed stacking an enhanced Swin Transformer, improved ConvNeXt, spatial attention mechanism. transformers incorporate two key components- hierarchical feature extraction shifting window mechanism extract from garbage images effectively. extracts most important regions identify In contrast, captures long-range dependencies within image garbage. ConvNext block optimized parameterization local image. This capability enables model discern fine-grained details individual particles, shape, texture, subtle variations color appearance, leading more accurate classification results. When evaluated performance using publicly available Garbage Classification dataset, it attained 98.97% accuracy, 98.42% Precision, 98.61% Recall. Due its lightweight low time power, surpasses existing state-of-the-art models.

Language: Английский

---

Revolutionizing Industrial Park Waste Classification with Artificial Intelligence: A Behavioral Economics and Evolutionary Game Theory Perspective

Process Safety and Environmental Protection, Journal Year: 2025, Volume and Issue: unknown, P. 107043 - 107043

Published: March 1, 2025

Language: Английский

---

ПОЛИМЕРЛІ КОНТЕЙНЕРЛЕРДІ СҰРЫПТАУ ПРОЦЕСІН ОҢТАЙЛАНДЫРУ ҮШІН НЕЙРОНДЫҚ ЖЕЛІЛЕРДІ ҚОЛДАНУ

Вестник Академии гражданской авиации, Journal Year: 2025, Volume and Issue: 36(1)

Published: March 1, 2025

Бұл зерттеуде пластикалық контейнерлерді тиімді сұрыптау үшін конволюционды нейрондық желілерді (CNN) және ұзақ қысқа мерзімді жадты (LSTM) біріктіретін гибридті желі архитектурасын пайдалануды қарастырады. Зерттеу жақын инфрақызыл (NIR) спектроскопиялық құрылғысымен алынған, химиялық құрамы мен ластану деңгейіне байланысты қалдықтарды жіктеуге бағытталған. Эксперимент нәтижелері CNN+LSTM моделі пластиктердің әртүрлі түрлері түстерін тану, соның ішінде контейнерлердегі ластаушы заттарды анықтауда салыстырмалы түрде жоғары дәлдікке қол жеткізетінін көрсетеді. Модельдің өнімділігін бағалау логистикалық регрессия, ішінара ең кіші квадраттар (PLS) сызықтық дискриминантты талдау (LDA) сияқты дәстүрлі жіктеу әдістерімен жүргізілді. Нәтижелер үлгісі тәсілдерге қарағанда, әсіресе класстар арасындағы спектрлік айырмашылықтары аз сценарийлерде тиімдірек жұмыс істейтінін зерттеу қайта өңдеу процестерінің тиімділігін арттыру машиналық оқытудың әлеуетін көрсетеді, осылайша экологиялық тұрақтылықты жақсартуға ықпал етеді.

---

Optimizing waste management with integrated AIoT, edge computing, and LoRaWAN communication technologies

Internet of Things, Journal Year: 2025, Volume and Issue: unknown, P. 101546 - 101546

Published: March 1, 2025

Language: Английский

---

Weakly Supervised Waste Classification with Adaptive Loss and Enhanced Class Activation Maps

Lecture notes in computer science, Journal Year: 2025, Volume and Issue: unknown, P. 347 - 361

Published: Jan. 1, 2025

Language: Английский

---

Smart waste management and classification system using advanced IoT and AI technologies

PeerJ Computer Science, Journal Year: 2025, Volume and Issue: 11, P. e2777 - e2777

Published: April 1, 2025

The effective management of municipal solid waste is a critical global issue, affecting both urban and rural areas. To address the growing volume waste, proactive planning essential. Traditionally, often disposed without segregation, preventing recycling recovery raw materials. Proper segregation fundamental requirement for management, allowing materials to be recycled efficiently. Emerging technologies such as artificial intelligence (AI), machine learning (ML), Internet Things (IoT) offer powerful tools identifying recyclable like glass, plastic, metal within waste. primary goal this research contribute cleaner environment, reduce infant mortality, improve maternal health, support efforts combat HIV/AIDS, malaria, other diseases. This study introduces an intelligent smart system (iSSWMs) designed smartly collect segregate proposed focuses on three types materials: metal. first phase involves collection using bins connected mobile application, which sends notifications when are full. In second phase, we develop deep learning-based mechanical model VGG-19 model, achieved performance accuracy 99.7% during training. best our knowledge, iSSWMs promising framework that integrates through use cutting-edge technologies, delivering high efficiency.

Language: Английский

---