A knowledge‐based materials descriptor for compositional dependence of phase transformation in NiTi shape memory alloys DOI Creative Commons
Cheng Li,

Qingkai Liang,

Yumei Zhou

et al.

Materials Genome Engineering Advances, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 11, 2024

Abstract This study presents ∆τ, a novel descriptor that captures the compositional dependence of phase transformation temperature (Ap) in NiTi‐based shape memory alloys (SMAs). Designed to address complexity multicomponent SMAs, ∆τ was integrated into symbolic regression (SR) and kernel ridge (KRR) models, yielding substantial improvements predicting key functional properties: temperature, enthalpy, thermal hysteresis. Using KRR model with we explored NiTiHfZrCu space, identifying six promising high Ap (>250°C), large enthalpy (>27 J/g), low Experimental validation confirmed model's accuracy showing high‐temperature behavior hysteresis, suitable for high‐performance applications aerospace nuclear industries. These findings underscore power domain‐informed descriptors like enhancing machine learning‐driven materials design.

Language: Английский

Enhancing the Spatial Resolution of Sentinel-2 Images Through Super-Resolution Using Transformer-Based Deep Learning Models DOI Creative Commons
Alireza Sharifi,

Mohammad Mahdi Safari

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2025, Volume and Issue: 18, P. 4805 - 4820

Published: Jan. 1, 2025

Satellite imagery plays a pivotal role in environmental monitoring, urban planning, and national security. However, spatial resolution limitations of current satellite sensors restrict the clarity usability captured images. This study introduces novel transformer-based deep learning model to enhance Sentinel-2 The proposed architecture leverages Multi-Head Attention integrated Spatial Channel mechanisms effectively extract reconstruct fine details from low-resolution inputs. model's performance was evaluated on dataset, along with benchmark datasets (AID UC-Merced), compared against state-of-the-art methods, including ResNet, Swin Transformer, ViT. Experimental results demonstrate superior performance, achieving PSNR 33.52 dB, SSIM 0.862, SRE 36.7 dB RGB bands. method outperforms approaches, ViT, (Sentinel-2, AID, UC-Merced.The that achieves terms PSNR, SSIM, metrics, highlighting its effectiveness revealing finer improving image quality for practical remote sensing applications.

Language: Английский

Citations

3
Regional soil water content monitoring based on time-frequency spectrogram of low-frequency swept acoustic signal DOI Creative Commons

Kangle Song,

Jing Nie, Yang Li

et al.

Geoderma, Journal Year: 2024, Volume and Issue: 441, P. 116765 - 116765

Published: Jan. 1, 2024

Acoustic waves offer a non-destructive, safe, and cost-effective means of monitoring the environment, with potential application in soil water content monitoring. However, extracting information from acoustic signals is still challenging. To tackle this issue, we have developed low-frequency swept signal detection device system. We conducted penetration testing using signals. The swept-frequency passing through were transformed into time–frequency spectrogram. Using Swin-Transformer model, established regression model between spectrogram frequencies content. Predictions made both on laboratory test dataset field trials calibrated model. results indicate that RMSE, MAE, R2 values observed model's outputs (%) for are 0.191, 0.081, 0.999, respectively, In case trials, predicted 6.715 %, 1.829 0.711, respectively. These studies demonstrate method highly effective predicting content, best achieved at resolution 20 PPI (Pixels Per Inch) within frequency range 260–360 Hz. It provides an efficient approach detection, effectively resolves difficulty building models caused by single-parameter limitation traditional

Language: Английский

Citations

10
Optimizing the sustainable performance of public buildings: A hybrid machine learning algorithm DOI
Wen Xu,

Xianguo Wu,

Shishu Xiong

et al.

Energy, Journal Year: 2025, Volume and Issue: unknown, P. 135283 - 135283

Published: Feb. 1, 2025

Language: Английский

Citations

1
Recognition of mango and location of picking point on stem based on a multi-task CNN model named YOLOMS DOI
Bin Zhang, Yuyang Xia, Rongrong Wang

et al.

Precision Agriculture, Journal Year: 2024, Volume and Issue: 25(3), P. 1454 - 1476

Published: Feb. 5, 2024

Language: Английский

Citations

7
UAVEC-FLchain: Distributed multi-regional jujube orchard joint yield estimation for secure agricultural-IoT applications DOI
Jing Nie, Jiachen Jiang, Yang Li

et al.

Internet of Things, Journal Year: 2024, Volume and Issue: 25, P. 101143 - 101143

Published: Feb. 28, 2024

Language: Английский

Citations

6
Accurate Recognition of Jujube Tree Trunks Based on Contrast Limited Adaptive Histogram Equalization Image Enhancement and Improved YOLOv8 DOI Open Access
Shunkang Ling, Nianyi Wang, Jingbin Li

et al.

Forests, Journal Year: 2024, Volume and Issue: 15(4), P. 625 - 625

Published: March 29, 2024

The accurate recognition of tree trunks is a prerequisite for precision orchard yield estimation. Facing the practical problems complex environment and large data flow, existing object detection schemes suffer from key issues such as poor quality, low timeliness accuracy, weak generalization ability. In this paper, an improved YOLOv8 designed on basis flow screening enhancement lightweight jujube trunk detection. Firstly, frame extraction algorithm was proposed utilized to efficiently screen effective data. Secondly, CLAHE image method used enhance quality. Finally, backbone model replaced with GhostNetv2 structure transformation, also introducing CA_H attention mechanism. Extensive comparison ablation results show that average quality-enhanced dataset over original increases 81.2% 90.1%, YOLOv8s-GhostNetv2-CA_H in paper reduces size by 19.5% compared YOLOv8s base model, increasing 2.4% 92.3%, recall 1.4%, [email protected] 1.8%, FPS being 17.1% faster.

Language: Английский

Citations

5
Fuzzy EfficientDet: An Approach for Precise Detection of Larch Infestation Severity in UAV Imagery Under Dynamic Environmental Conditions DOI Creative Commons
Shuo Yang, Jingbin Li, Yang Li

et al.

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2024, Volume and Issue: 17, P. 8810 - 8822

Published: Jan. 1, 2024

In this paper, a novel deep learning framework, fuzzy EfficientDet, is proposed to address the challenge of accurately detecting larch infested by Coleophora laricella pests in UAV imagery, where key innovation incorporation spatial attention mechanism (FSAM), which can effectively deal with problem model uncertainty due complexity environmental transformations and image features. First, study designs implements Global-Local Squeeze-and-Excitation Module, profoundly integrates global local feature information, realizes dynamic adaptation importance channels EfficientNet, thus improves overall expression efficiency network. Second, constructed dense Bi-FPN architecture, adds connection structure original enhance modeling accuracy for small targets long-range dependencies. Finally, develops mitigate unstable performance EfficientDet face fluctuations triggered changes lighting conditions seasonal effects. Experiments demonstrate that shows superior compared traditional SSD, Faster R-CNN, YOLO V5, unimproved target detection method on Swedish Forest Agency (2021) dataset, its mAP as high 94.29%. This result demonstrates provides an efficient reliable solution when dealing task images, especially complex extraction.

Language: Английский

Citations

4
Dynamic risk early warning system for tunnel construction based on two-dimensional cloud model DOI

Huaiyuan Sun,

Mengqi Zhu,

Yiming Dai

et al.

Expert Systems with Applications, Journal Year: 2024, Volume and Issue: 255, P. 124799 - 124799

Published: July 14, 2024

Language: Английский

Citations

4
Brain-inspired artificial intelligence research: A review DOI Creative Commons
Guoyin Wang, Huanan Bao, Qun Liu

et al.

Science China Technological Sciences, Journal Year: 2024, Volume and Issue: 67(8), P. 2282 - 2296

Published: July 30, 2024

Artificial intelligence (AI) systems surpass certain human abilities in a statistical sense as whole, but are not yet the true realization of these and behaviors. There differences, even contradictions, between cognition behavior AI humans. With goal achieving general AI, this study contains review role cognitive science inspiring development three mainstream academic branches based on three-layer framework proposed by David Marr, limitations current explored analyzed. The differences inconsistencies mechanisms brain computation They found to be cause contradictions Additionally, eight important research directions their scientific issues that need focus brain-inspired proposed: highly imitated bionic information processing, large-scale deep learning model balances structure function, multi-granularity joint problem solving bidirectionally driven data knowledge, models simulate specific structures, collaborative processing mechanism with physical separation perceptual interpretive analysis, embodied integrates mechanisms, simulation from individual group (social intelligence), AI-assisted intelligence.

Language: Английский

Citations

4
Cotton Area Extraction Based on High-Resolution Sentinel-2 Satellite Images DOI Creative Commons
Xihuizi Liang,

Haiyan Tian,

Wen Wang

et al.

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2025, Volume and Issue: 18, P. 10924 - 10936

Published: Jan. 1, 2025

Language: Английский

Citations

0