A Real-Time Green and Lightweight Model for Detection of Liquefied Petroleum Gas Cylinder Surface Defects Based on YOLOv5

Applied Sciences, Год журнала: 2025, Номер 15(1), С. 458 - 458

Опубликована: Янв. 6, 2025

Industry requires defect detection to ensure the quality and safety of products. In resource-constrained devices, real-time speed, accuracy, computational efficiency are most critical requirements for detection. This paper presents a novel approach surface defects on LPG cylinders, utilising an enhanced YOLOv5 architecture referred as GLDD-YOLOv5. The integrates ghost convolution ECA blocks improve feature extraction with less overhead in network’s backbone. It also modifies P3–P4 head structure increase speed. These changes enable model focus more effectively small medium-sized defects. Based comparative analysis other YOLO models, proposed method demonstrates superior performance. Compared base YOLOv5s model, achieved 4.6% average 44% reduction cost, 45% decrease parameter counts, 26% file size. experimental evaluations RTX2080Ti, inference rate 163.9 FPS total carbon footprint 0.549 × 10−3 gCO2e. technique offers efficient robust eco-friendly solution compatible edge computing devices.

Язык: Английский

---

YOLO-SAG: An improved wildlife object detection algorithm based on YOLOv8n

Ecological Informatics, Год журнала: 2024, Номер 83, С. 102791 - 102791

Опубликована: Авг. 23, 2024

Язык: Английский

---

A Comparative Analysis of YOLOv9, YOLOv10, YOLOv11 for Smoke and Fire Detection

Fire, Год журнала: 2025, Номер 8(1), С. 26 - 26

Опубликована: Янв. 13, 2025

Forest fires cause extensive environmental damage, making early detection crucial for protecting both nature and communities. Advanced computer vision techniques can be used to detect smoke fire. However, accurate of fire in forests is challenging due different factors such as shapes, changing light, similarity with other smoke-like elements clouds. This study explores recent YOLO (You Only Look Once) deep-learning object models YOLOv9, YOLOv10, YOLOv11 detecting forest environments. The evaluation focuses on key performance metrics, including precision, recall, F1-score, mean average precision (mAP), utilizes two benchmark datasets featuring diverse instances across findings highlight the effectiveness small version (YOLOv9t, YOLOv10n, YOLOv11n) tasks. Among these, YOLOv11n demonstrated highest performance, achieving a 0.845, recall 0.801, mAP@50 0.859, mAP@50-95 0.558. versions (YOLOv11n YOLOv11x) were evaluated compared against several studies that employed same datasets. results show YOLOv11x delivers promising variants models.

Язык: Английский

---

Optimizing Solar Panel Classification with Yolov11: Bridging Physics Principles and Artificial Intelligence for Enhanced Performance

African Journal of Biomedical Research, Год журнала: 2025, Номер unknown, С. 80 - 91

Опубликована: Янв. 6, 2025

This paper presents a novel framework for solar panel classification, leveraging physics-informed enhancements integrated into the YOLOv11 architecture. By incorporating domain-specific augmentations such as tilt-induced irradiance adjustments, shading simulations, and temperature effects, model demonstrates significant improvements in performance robustness. A comprehensive dataset of over 10,000 high-resolution images was created, encompassing diverse environmental conditions, tilt angles, levels to replicate real-world scenarios. Physics-informed resulted 7.3% increase mean average precision (mAP) 12% improvement accuracy under challenging extreme occlusions, compared traditional methods. The optimized achieved top-1 91%, an mAP 89.7%, inference speed 25 FPS. study highlights integration physics-based insights deep learning pipelines transformative approach analysis, paving way more reliable scalable renewable energy monitoring systems.

Язык: Английский

---

Research on tea buds detection based on optimized YOLOv5s

IET Image Processing, Год журнала: 2025, Номер 19(1)

Опубликована: Янв. 1, 2025

Abstract As one of the world's most popular beverages, tea plays a significant role in improving production efficiency and quality through identification shoots during manufacturing process. However, due to complex morphology, small size, susceptibility factors like lighting obstruction, traditional methods suffer from low accuracy efficiency. In this study, image enhancement techniques such as HSV transformation, horizontal flipping, vertical flipping were applied training dataset improve model robustness enhance generalization across varying angles. To address these challenges context buds detection, deep‐learning‐based object detection have emerged promising solutions. Nevertheless, current technologies still face limitations when detecting under conditions. performance, article proposed an improved YOLOv5s (You Only Look Once version 5 model) algorithm. algorithm, CBAM, SE, CA attention mechanisms incorporated into backbone network augment feature extraction, weighted Bidirectional Feature Pyramid Network (BiFPN) is employed neck boost resulting YOLOv5s_teabuds model. Experimental results indicated that significantly outperformed original terms precision, recall, mAP F1‐score, with mechanism providing notable improvement—enhancing F1‐score by 18.119%, 9.633%, 16.496% 13.524%, respectively. After integrating BiFPN, further strengthened performance robustness, increased 19.346%, 11.388%, 18.620%, 15.059%, prove optimized can provide real‐time, high‐precision method for robotic harvesting.

Язык: Английский

---

Evaluation of the Performance of a YOLOv10-Based Deep Learning Model for Tooth Detection and Numbering on Panoramic Radiographs of Patients in the Mixed Dentition Period

Diagnostics, Год журнала: 2025, Номер 15(4), С. 405 - 405

Опубликована: Фев. 7, 2025

Objectives: This study evaluated the performance of a YOLOv10-based deep learning model in detecting and numbering teeth panoramic radiographs pediatric patients mixed dentition period. Methods: Panoramic radiographic images from 200 period, each with at least 10 primary underlying permanent tooth germs, were included study. A total 8153 these manually labeled. The dataset was divided for development artificial intelligence model, 70% used training, 15% testing, validation. Results: precision, recall, mAP50, mAP50-95, F1 score detection found to be 0.90, 0.94, 0.968, 0.696, 0.919, respectively. Conclusions: models can accurately detect number which support clinicians their daily practice. Future works may focus on optimization across varied cases enhance clinical applicability.

Язык: Английский

---

Triple-stage crack detection in stone masonry using YOLO-ensemble, MobileNetV2U-net, and spectral clustering

Automation in Construction, Год журнала: 2025, Номер 172, С. 106045 - 106045

Опубликована: Фев. 7, 2025

Язык: Английский

---

LW-UAV–YOLOv10: A lightweight model for small UAV detection on infrared data based on YOLOv10

GEOMATICA, Год журнала: 2025, Номер 77(1), С. 100049 - 100049

Опубликована: Фев. 7, 2025

Язык: Английский

---

An Enhanced Algorithm for Detecting Small Traffic Signs Using YOLOv10

Electronics, Год журнала: 2025, Номер 14(5), С. 955 - 955

Опубликована: Фев. 27, 2025

Recognizing traffic signs is crucial for autonomous driving systems, as it significantly impacts their safety and dependability. However, challenges like the diminutive size of objects intricate background environments limit effectiveness current object detection models. To improve small sign detection, this research introduces an enhanced algorithm built on YOLOv10. First, a custom-designed layer detecting integrated into neck section network, enhancing feature extraction process these objects. Second, refined downsampling module, called Triple-Branch Downsampling (TBD), utilizes multi-branch structure hybrid pooling strategy to boost efficiency within model. Finally, loss function optimized by integrating Normalized Wasserstein Distance (NWD) Wise-MPDIoU mechanisms, increasing accuracy bounding box matching regression. The experimental findings indicate that reaches [email protected] 84.8%, marking 4% increase over classification recall are 73.4% 82.9%, respectively. Moreover, parameter count decreases approximately 10%, while computational complexity reduced around 5%.

Язык: Английский

---

Detection of Small-Sized Electronics Endangering Facilities Involved in Recycling Processes Using Deep Learning

Applied Sciences, Год журнала: 2025, Номер 15(5), С. 2835 - 2835

Опубликована: Март 6, 2025

In Japan, local governments implore residents to remove the batteries from small-sized electronics before recycling them, but some products still contain lithium-ion batteries. These residual may cause fires, resulting in serious injuries or property damage. Explosive materials such as mobile (such power banks) have been identified fire investigations. Therefore, these fire-causing items should be detected and separated regardless of whether other processes are use. This study focuses on automatic detection using deep learning electronic products. Mobile were chosen first target this approach. study, MATLAB R2024b was applied construct You Only Look Once version 4 algorithm. The model trained enable results show that model’s average precision value reached 0.996. Then, expanded three categories items, including batteries, heated tobacco (electronic cigarettes), smartphones. Furthermore, real-time object videos detector carried out. able detect all accurately. conclusion, technologies significant promise a method for safe high-quality recycling.

Язык: Английский

---