A Comparative Analysis of Deep-Learning-Based YOLO Models (V8n and V8s) for Object Detection Using GSV Images DOI

Arenla Longchar,

Misal Digvijay Anna,

Rajesh K. Dhumal

et al.

Published: Nov. 1, 2023

This study presents a comparative analysis of deep-learning-based YOLO (You Only Look Once) models, namely YOLOv8n and YOLOv8s, for detecting power poles along street in Nellore, Andhra Pradesh. The objective is to assess how well efficiently these models accurately detect Google Street View (GSV) images. utilizes dataset consisting view images that are annotated used training the YOLOv8s which then tested on set different To verify models' accuracy effectiveness recognizing poles, evaluation criteria like precision, recall, F1 score used. results indicate both effective Nellore. However, model has greater 81% compared model's 78%. findings this work demonstrate potential pole detection using GSV offers valuable insights researchers individuals GIS computer vision field, contributing development efficient accurate methods infrastructure monitoring management.

Language: Английский

Definition of a reference standard for performance evaluation of autonomous vehicles real-time obstacle detection and distance estimation in complex environments DOI
Tabinda Naz Syed, Jun Zhou, Francesco Marinello

et al.

Computers and Electronics in Agriculture, Journal Year: 2025, Volume and Issue: 232, P. 110143 - 110143

Published: Feb. 21, 2025

Language: Английский

Citations

3
CURI-YOLOv7: A Lightweight YOLOv7tiny Target Detector for Citrus Trees from UAV Remote Sensing Imagery Based on Embedded Device DOI Creative Commons
Yali Zhang,

Xipeng Fang,

Jun Guo

et al.

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(19), P. 4647 - 4647

Published: Sept. 22, 2023

Data processing of low-altitude remote sensing visible images from UAVs is one the hot research topics in precision agriculture aviation. In order to solve problems large model size with slow detection speed that lead inability process real time, this paper proposes a lightweight target detector CURI-YOLOv7 based on YOLOv7tiny which suitable for individual citrus tree UAV imagery. This augmented dataset morphological changes and Mosica Mixup. A backbone depthwise separable convolution MobileOne-block module was designed replace YOLOv7tiny. SPPF (spatial pyramid pooling fast) used original spatial structure. Additionally, we redesigned neck by adding GSConv depth-separable deleted its input layer (80, 80) output head 80). new ELAN structure designed, redundant convolutional layers were deleted. The experimental results show GFLOPs = 1.976, parameters 1.018 M, weights 3.98 MB, mAP 90.34% imagery trees dataset. single image 128.83 computer 27.01 embedded devices. Therefore, can basically achieve function forms foundation subsequent real-time identification geographic coordinates positioning, conducive study precise agricultural management orchards.

Language: Английский

Citations

13
Automatic Counting and Location Labeling of Rice Seedlings from Unmanned Aerial Vehicle Images DOI Open Access
Jui‐Feng Yeh,

Kuei-Mei Lin,

Li-Ching Yuan

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(2), P. 273 - 273

Published: Jan. 8, 2024

Traditional counting of rice seedlings in agriculture is often labor-intensive, time-consuming, and prone to errors. Therefore, agricultural automation has gradually become a prominent solution. In this paper, UVA detection, combining deep learning with unmanned aerial vehicle (UAV) sensors, contributes precision agriculture. We propose YOLOv4-based approach for the location marking from images. The detection tiny objects crucial challenging task imagery. we make modifications data augmentation activation functions neural elements model meet requirements seedling counting. preprocessing stage, segment UAV images into different sizes training. Mish employed enhance accuracy YOLO one-stage detector. utilize dataset provided AIdea 2021 competition evaluate system, achieving an F1-score 0.91. These results indicate superiority proposed method over baseline system. Furthermore, outcomes affirm potential precise

Language: Английский

Citations

4
Parallel RepConv network: Efficient vineyard obstacle detection with adaptability to multi-illumination conditions DOI

Xuezhi Cui,

Licheng Zhu, Bo Zhao

et al.

Computers and Electronics in Agriculture, Journal Year: 2025, Volume and Issue: 230, P. 109901 - 109901

Published: Jan. 8, 2025

Language: Английский

Citations

0
Vision-based power line cables and pylons detection for low flying aircraft DOI Creative Commons

Jakub Gwizdała,

Doruk Oner,

Soumava Kumar Roy

et al.

Machine Vision and Applications, Journal Year: 2025, Volume and Issue: 36(2)

Published: Feb. 10, 2025

Language: Английский

Citations

0
Defect Detection Algorithm for Battery Cell Casings Based on Dual-Coordinate Attention and Small Object Loss Feedback DOI Open Access
Tianjian Li,

Jiale Ren,

Qingping Yang

et al.

Processes, Journal Year: 2024, Volume and Issue: 12(3), P. 601 - 601

Published: March 18, 2024

To address the issue of low accuracy in detecting defects battery cell casings with space ratio and small object characteristics, feature are studied, an detection algorithm based on dual-coordinate attention loss feedback is proposed. Firstly, EfficientNet-B1 backbone network employed for extraction. Secondly, a module introduced to preserve more positional information through dual branches embed into channel precise localization features. Finally, incorporated after bidirectional pyramid (BiFPN) fusion, balancing contribution overall loss. Experimental comparisons casing dataset demonstrate that proposed outperforms EfficientDet-D1 algorithm, average precision improvement 4.23%. Specifically, scratches features, 13.21%; wrinkles 9.35%; holes 3.81%. Moreover, time 47.6 ms meets requirements practical production.

Language: Английский

Citations

2
Artificial Intelligence in Aviation Safety: Systematic Review and Biometric Analysis DOI Creative Commons
Gülay Demir, Sarbast Moslem, Szabolcs Duleba

et al.

International Journal of Computational Intelligence Systems, Journal Year: 2024, Volume and Issue: 17(1)

Published: Nov. 12, 2024

This study aims to offer aviation safety researchers, practitioners, and decision-makers a comprehensive exploration of integrating advanced technologies, such as artificial intelligence machine learning, inform fortify future strategies. Focusing on systematic bibliometric perspectives, the paper reviewed 224 articles in Scopus database from 2004 2024 (January). Key findings highlight China's notable contributions research, underscoring its leadership international collaboration. The techniques employed encompass time series models, deep AI, neurophysiological modeling, optimization algorithms. analysis discerns prominent research trends, including accident analysis, pilot behavior, measures, endeavors enhance standards. industry's steadfast commitment safety, efficiency, technological innovation is evident. By uncovering main structures, foci, trends this equips researchers practitioners with crucial insights into ongoing potential developments, fostering more profound understanding safety.

Language: Английский

Citations

2
New Trends and Challenges in Precision and Digital Agriculture DOI Creative Commons
Gniewko Niedbała, Magdalena Piekutowska, Patryk Hara

et al.

Agronomy, Journal Year: 2023, Volume and Issue: 13(8), P. 2136 - 2136

Published: Aug. 15, 2023

Real change is needed in the agricultural sector to meet challenges of 21st century terms humanity’s food needs [...]

Language: Английский

Citations

5
Carbonate reservoir fracture‐cavity system identification based on the improved YOLOv5s deep learning algorithm DOI Creative Commons

Xiaoyong Feng,

Kai Zhao, Jianguo Zhang

et al.

Energy Science & Engineering, Journal Year: 2024, Volume and Issue: 12(6), P. 2643 - 2660

Published: May 27, 2024

Abstract In carbonate reservoirs characterized by the fracture‐cavity system as storage spaces, drilling process is highly prone to loss of fluid. This not only affects efficiency but can also lead severe accidents, such blowouts. Therefore, it crucial understand distribution pattern these fractures. However, formation rock systems, being controlled various factors, difficult precisely identify. limitation hampers efficient development types oil and gas fields. paper presents a case study M5 5 sub‐section reservoir in Sulige gasfield, proposing an improved You Only Look Once v5s (YOLOv5s) deep learning algorithm. It utilizes enhanced training with conventional logging data identify response characteristics fractures reservoirs. And its identification results have been confirmed be accurate fracture obtained through different means, core samples, cast thin section photographs, imaging data, seismic attributes. method incorporates Ghost convolution module replace Conv backbone network YOLOv5s model, modifies C3 into Bottleneck module, effectively making model more lightweight. Additionally, Convolutional Block Attention Module integrated Neck network, enhancing model's feature extraction capabilities. Finally, employs Efficient Intersection over Union Loss function instead Complete Loss, reducing network's regression loss. The validation using actual demonstrate that this achieves average recognition accuracy 87.3% for system, which 3% improvement baseline (YOLOv5s). enhancement beneficial locating fluid positions

Language: Английский

Citations

1
Experimental Study on the Peeling Fracture Effect of Fresh Corn Ear Based on High and Low Roller Peeling Equipment DOI Creative Commons
Shun Chen, Xinwei Zhang,

Chunxia Jiang

et al.

Agriculture, Journal Year: 2023, Volume and Issue: 13(8), P. 1585 - 1585

Published: Aug. 9, 2023

Aiming to address the problems of low working efficiency and high damage rate roller peeling equipment in process fresh corn harvesting China, this paper theoretically analyzes mechanical motion between device ear, a high–low roll structure is proposed. This incorporates elastomeric rubber material, segmentation design, an adjustable spiral frame, selection relevant parameters given. To determine optimal operating for fresh-corn-peeling device, three-factor, three-level orthogonal test was conducted using Box–Behnken central grouping method Design-Expert 12 software. The factors were speed, tilt angle, vibrating plate frequency. evaluation indices considered bract (BPR) grain breaking (GBR). Based on theoretical analysis results, bench fresh-corn-ear-peeling established parameter combination quality determined according actual work situation. results show that impact BPR GBR, from large small, following order: frequency vibration plate. optimization module used optimize integers obtain combination: speed 480 r·min−1; angle 8°; 260 times·min−1; corresponding 91.75%, which 0.66% points lower than value; GBR 1.55%, 0.08% higher value. Notably, exhibited superior performance terms fracture compared with standard equipment. Therefore, study provides valuable technical support design

Language: Английский

Citations

1