Deep Learning for Sustainable Agriculture: A Systematic Review on Applications in Lettuce Cultivation

Sustainability, Год журнала: 2025, Номер 17(7), С. 3190 - 3190

Опубликована: Апрель 3, 2025

Lettuce, a vital economic crop, benefits significantly from intelligent advancements in its production, which are crucial for sustainable agriculture. Deep learning, core technology smart agriculture, has revolutionized the lettuce industry through powerful computer vision techniques like convolutional neural networks (CNNs) and YOLO-based models. This review systematically examines deep learning applications including pest disease diagnosis, precision spraying, pesticide residue detection, crop condition monitoring, growth stage classification, yield prediction, weed management, irrigation fertilization management. Notwithstanding significant contributions, several critical challenges persist, constrained model generalizability dynamic settings, exorbitant computational requirements, paucity of meticulously annotated datasets. Addressing these is essential improving efficiency, adaptability, sustainability learning-driven solutions production. By enhancing resource reducing chemical inputs, optimizing cultivation practices, contributes to broader goal explores research progress, optimization strategies, future directions strengthen learning’s role fostering farming.

Язык: Английский

---

Optimizing Deep Learning Acceleration on FPGA for Real-Time and Resource-Efficient Image Classification

Applied Sciences, Год журнала: 2025, Номер 15(1), С. 422 - 422

Опубликована: Янв. 5, 2025

Deep learning (DL) has revolutionized image classification, yet deploying convolutional neural networks (CNNs) on edge devices for real-time applications remains a significant challenge due to constraints in computation, memory, and power efficiency. This work presents an optimized implementation of VGG16 VGG19, two widely used CNN architectures, classifying the CIFAR-10 dataset using transfer field-programmable gate arrays (FPGAs). Utilizing Xilinx Vitis-AI TensorFlow2 frameworks, we adapt VGG19 FPGA deployment through quantization, compression, hardware-specific optimizations. Our achieves high classification accuracy, with Top-1 accuracy 89.54% 87.47% respectively, while delivering reductions inference latency (7.29× 6.6× compared CPU-based alternatives). These results highlight suitability our approach resource-efficient, applications. Key contributions include detailed methodology combining acceleration, analysis hardware resource utilization, performance benchmarks. underscores potential FPGA-based solutions enable scalable, low-latency DL deployments domains such as autonomous systems, IoT, mobile devices.

Язык: Английский

---

Deep learning based image classification for embedded devices: A systematic review

Neurocomputing, Год журнала: 2025, Номер 623, С. 129402 - 129402

Опубликована: Янв. 15, 2025

Язык: Английский

---

Structural Health Monitoring of Laminated Composites Using Lightweight Transfer Learning

Machines, Год журнала: 2024, Номер 12(9), С. 589 - 589

Опубликована: Авг. 25, 2024

Due to their excellent strength-to-weight ratio, composite laminates are gradually being substituted for traditional metallic materials in a variety of industries. However, due orthotropic nature, prone several different types damage, with delamination the most prevalent and serious. Therefore, deep learning-based methods that use sensor data conduct autonomous health monitoring have drawn much interest structural (SHM). direct application these models is restricted by lack training data, necessitating transfer learning. The commonly used learning computationally expensive; therefore, present research proposes lightweight (LTL) SHM composites. an EfficientNet–based LTL model only requires fine-tuning target vibration rather than from scratch. Wavelet-transformed vibrational various classes utilized confirm effectiveness proposed method. Moreover, assessment measures applied assess performance on unseen test datasets. outcomes validation show pre-trained could successfully perform laminates, achieving high values regarding accuracy, precision, recall, F1-score.

Язык: Английский

---

Optimizing RAG Techniques for Automotive Industry PDF Chatbots: A Case Study with Locally Deployed Ollama ModelsOptimizing RAG Techniques Based on Locally Deployed Ollama ModelsA Case Study with Locally Deployed Ollama Models

Опубликована: Окт. 25, 2024

Язык: Английский

---

Lightweight Speaker Verification with Integrated VAD and Speech Enhancement

Digital Signal Processing, Год журнала: 2025, Номер unknown, С. 104969 - 104969

Опубликована: Янв. 1, 2025

Язык: Английский

---

Machine Learning-Based Resource Management in Fog Computing: A Systematic Literature Review

Sensors, Год журнала: 2025, Номер 25(3), С. 687 - 687

Опубликована: Янв. 23, 2025

This systematic literature review analyzes machine learning (ML)-based techniques for resource management in fog computing. Utilizing the Preferred Reporting Items Systematic Reviews and Meta-Analyses (PRISMA) protocol, this paper focuses on ML deep (DL) solutions. Resource computing domain was thoroughly analyzed by identifying key factors constraints. A total of 68 research papers extended versions were finally selected included study. The findings highlight a strong preference DL addressing challenges within paradigm, i.e., 66% reviewed articles leveraged techniques, while 34% utilized ML. Key such as latency, energy consumption, task scheduling, QoS are interconnected critical optimization. analysis reveals that prime addressed ML-based management. Latency is most frequently parameter, investigated 77% articles, followed consumption scheduling at 44% 33%, respectively. Furthermore, according to our evaluation, an extensive range challenges, computational scalability management, data availability quality, model complexity interpretability, employing 73, 53, 45, 46 ML/DL

Язык: Английский

---

Resource-Constrained Binary Image Classification

Lecture notes in computer science, Год журнала: 2025, Номер unknown, С. 215 - 230

Опубликована: Янв. 1, 2025

Язык: Английский

---

Resource constraint crop damage classification using depth channel shuffling

Engineering Applications of Artificial Intelligence, Год журнала: 2025, Номер 144, С. 110117 - 110117

Опубликована: Янв. 29, 2025

Язык: Английский

---

Resource-Efficient Clustered Federated Learning Framework for Industry 4.0 Edge Devices

AI, Год журнала: 2025, Номер 6(2), С. 30 - 30

Опубликована: Фев. 6, 2025

Industry 4.0 is an aggregate of recent technologies including artificial intelligence, big data, edge computing, and the Internet Things (IoT) to enhance efficiency real-time decision-making. data analytics demands a privacy-focused approach, federated learning offers viable solution for such scenarios. It allows each device train model locally using its own collected shares only updates with server without need share real data. However, communication computational costs sharing performance are major bottlenecks resource-constrained devices. This study introduces representative-based parameter-sharing framework that aims in environment. The begins by distributing initial devices, which then it send updated parameters back aggregation. To reduce costs, identifies groups devices similar parameter distributions sends from resourceful better-performing device, termed cluster head, server. A backup head also elected ensure reliability. Clustering performed based on device’s characteristics. Moreover, incorporates randomly selected past aggregated into current aggregation process through weighted averaging where more given greater weight performance. Comparative experimental evaluation state art testbed dataset demonstrates promising results minimizing cost while preserving prediction performance, ultimately enhances industrial environments.

Язык: Английский

---