A Cloud Computing Framework for Space Farming Data Analysis

AgriEngineering, Journal Year: 2025, Volume and Issue: 7(5), P. 149 - 149

Published: May 8, 2025

This study presents a system framework by which cloud resources are utilized to analyze crop germination status in 2U CubeSat. research aims address the onboard computing constraints nanosatellite missions boost space agricultural practices. Through Espressif Simple Protocol for Network-on-Wireless (ESP-NOW) technology, communication between ESP-32 modules were established. The corresponding sensor readings and image data securely streamed through Amazon Web Service Internet of Things (AWS IoT) an ESP-NOW receiver Roboflow. Real-time plant growth predictor monitoring was implemented web application provisioned at end. On other hand, sprouts on bed determined custom-trained Roboflow computer vision model. feasibility remote computational analysis CubeSat, given its minute form factor, successfully demonstrated proposed framework. detection model resulted mean average precision (mAP), precision, recall 99.5%, 99.9%, 100.0%, respectively. temperature, humidity, heat index, LED Fogger states, shown real time dashboard. With this use case, immediate actions can be performed accordingly when abnormalities occur. scalability nature allows adaptation various crops support sustainable activities extreme environments such as farming.

Language: Английский

---

Tiny Machine Learning and On-Device Inference: A Survey of Applications, Challenges, and Future Directions

Sensors, Journal Year: 2025, Volume and Issue: 25(10), P. 3191 - 3191

Published: May 19, 2025

The growth in artificial intelligence and its applications has led to increased data processing inference requirements. Traditional cloud-based solutions are often used but may prove inadequate for requiring near-instantaneous response times. This review examines Tiny Machine Learning, also known as TinyML, an alternative inference. focuses on where transmission delays make traditional Internet of Things (IoT) approaches impractical, thus necessitating a solution that uses TinyML on-device study, which follows the PRISMA guidelines, covers TinyML’s use cases real-world by analyzing experimental studies synthesizing current research characteristics experiments, such machine learning techniques hardware experiments. identifies existing gaps well means address these gaps. findings suggest strong record usability offers advantages over inference, particularly environments with bandwidth constraints require rapid discusses implications performance future applications.

Language: Английский

---

Real-Time Dolphin Whistle Detection on Raspberry Pi Zero 2 W with a TFLite Convolutional Neural Network

Robotics, Journal Year: 2025, Volume and Issue: 14(5), P. 67 - 67

Published: May 19, 2025

The escalating conflict between cetaceans and fisheries underscores the need for efficient mitigation strategies that balance conservation priorities with economic viability. This study presents a TinyML-driven approach deploying an optimized Convolutional Neural Network (CNN) on Raspberry Pi Zero 2 W real-time detection of bottlenose dolphin whistles, leveraging spectrogram analysis to address acoustic monitoring challenges. Specifically, CNN model previously developed classifying dolphins’ vocalizations originally implemented TensorFlow was converted Lite (TFLite) architectural optimizations, reducing size by 76%. Both TFLite models were trained 22 h underwater recordings taken in controlled environments processed into 0.8 s segments (300 × 150 pixels). Despite size, maintained same accuracy as original (87.8% vs. 87.0%). Throughput latency evaluated varying thread allocation (1–8 threads), revealing best performance at 4 threads (quad-core alignment), achieving inference 120 ms sustained throughput 8 spectrograms/second. system demonstrated robustness continuous stress tests without failure, underscoring its reliability marine environments. work achieved critical computational efficiency fidelity (F1-score: 86.9%) quantized, multithreaded inference. These advancements enable low-cost devices cetacean presence detection, offering transformative potential bycatch reduction adaptive deterrence systems. bridges artificial intelligence innovation ecological stewardship, providing scalable framework machine learning resource-constrained settings while addressing urgent

Language: Английский

---

Grading and Classification of Tomatoes Using ESP32-CAM and Impulse Cloud based IoT Platform for Sustainability

Procedia Computer Science, Journal Year: 2025, Volume and Issue: 260, P. 938 - 946

Published: Jan. 1, 2025

Language: Английский

---

Compression-Accuracy Co-optimization Through Hardware-aware Neural Architecture Search for Vibration Damage Detection

IEEE Internet of Things Journal, Journal Year: 2024, Volume and Issue: 11(19), P. 31745 - 31757

Published: June 26, 2024

Internet-of-Things (IoT) is a key enabler for the transition to Automatic Structural Health Monitoring (ASHM) of technical facilities, thanks seamless flow data from multitude always connected devices. Current IoT-ASHM installations, however, face double challenge ensure high accuracy while meeting requirement minimal energy consumption. The paper tackles these issues deep-learning perspective, and describes an IoT-enabled monitoring approach based on distributed end-to-end deep neural network (DNN). architecture supports both compression damage detection. A low-end microcontroller hosts specific local DNN; hardware-aware neural-architecture search strategy rules optimization, in order satisfy resource constraints set by computing features extracted feed aggregating unit, which includes stacked global classification layer full-scale After proper quantization, designed models are eventually deployed wireless accelerometer sensor. Finally, cost-benefit analysis evaluates system's impact sensor autonomy. Experiments well-known dataset proved that proposed solution could achieve state-of-the-art scores (all metrics above 98.4%) with transmission cost (less than 53 B average); as compared conventional approaches, described yielded reduction three orders magnitude

Language: Английский

---

An Ultrasensitive Device with Embedded Phononic Crystals for the Detection and Localisation of Nonlinear Guided Waves

Published: Jan. 1, 2024

Language: Английский

---

Non-destructive Detection of Water Adulteration Level in Fresh Milk Based on Combination of Dielectric Spectrum Technology and Machine Learning Method

Journal of Food Composition and Analysis, Journal Year: 2024, Volume and Issue: unknown, P. 106807 - 106807

Published: Oct. 1, 2024

Language: Английский

---

Application of Generative Artificial Intelligence in Minimizing Cyber Attacks on Vehicular Networks

Procedia Computer Science, Journal Year: 2024, Volume and Issue: 251, P. 140 - 149

Published: Jan. 1, 2024

Language: Английский

---

A lightweight optimization framework for real-time object detector on the embedded GPU platform

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: June 20, 2024

At present, embedded GPU devices play a central role in carrying edge intelligence applications, especially object detection. However, efficiently deploying and inferring convolutional neural network (CNN)-based detection models on with small size, low power consumption, limited computing resources remains significant challenge. In this paper, we proposed lightweight optimization framework for YOLOv3 to improve its inference efficiency onto GPU-based device, case the Jetson Nano, which is low-cost entry-level artificial (AI) computer markets. First, replace Backbone of by Efficient-RepVGG, more suitable architecture. Secondly, efficient bidirectional feature pyramid designed further compress model. Finally, Optimized Couple Head compensate loss accuracy without affecting effect model lightweight. After improvement, saves 92.26% 92.10% parameters computation, increasing speed 9.95 times, while maintaining high accuracy. Compared other detectors, about 1.6 2 times faster similar accuracy; 3 6 percentage higher speed.

Language: Английский

---

TinyML enabled smart parking dynamic slots computing and license plate recognition

2022 International Wireless Communications and Mobile Computing (IWCMC), Journal Year: 2024, Volume and Issue: unknown, P. 1583 - 1588

Published: May 27, 2024

Language: Английский

---