GF-1 WFV satellite images based forest cover mapping in China supported by open land use/cover datasets DOI Creative Commons
Xueli Peng, Guojin He, Guizhou Wang

et al.

Scientific Data, Journal Year: 2024, Volume and Issue: 11(1)

Published: Dec. 18, 2024

The United Nations sustainable development agenda emphasizes the importance of forests. China's forests cover 5% world's forest area, significantly influencing global climate and ecology. In recent decades, have undergone notable changes. Accurate maps are crucial for understanding distribution, conducting ecological research management. However, there is a lack satisfying criteria. To this issue, study focuses on developing precise 16-m resolution map China. For purpose, we propose classification framework based weakly supervised deep learning prior knowledge from open datasets. Utilizing GF-1 WFV satellite images, generated in 2020 named FCM16. FCM16 evaluated using 136,385 sample points, achieving an overall accuracy 94.64 ± 0.12%, producer's 91.12 0.27% user's 87.31 0.34%. Additionally, was compared with existing forest-related datasets, demonstrating its reliability. general, effectively represents 2020, providing valuable resource social analysis.

Language: Английский

A novel weakly-supervised method based on the segment anything model for seamless transition from classification to segmentation: A case study in segmenting latent photovoltaic locations DOI Creative Commons
Ruiqing Yang, Guojin He, Ranyu Yin

et al.

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2024, Volume and Issue: 130, P. 103929 - 103929

Published: May 25, 2024

In the quest for large-scale photovoltaic (PV) panel extraction, substantial data volumes are essential, given demand sub-meter rooftop PV resolution. This requires concept of Latent Photovoltaic Locations (LPL) to reduce scope amount subsequent processing. order minimize manual annotation, a pioneering weakly-supervised framework is proposed, which capable generating pixel-level annotations segmentation based on image-level and provides two datasets required classification-then-segmentation strategy without more annotations. The strong noise-resistance Segment Anything Model (SAM) discovered in extremely difficult rough coarse pseudo-label refinement, which, after integrating probability updating mechanism, achieves seamless transition from scene classification semantic segmentation. resulting national LPL distribution map, rendered at 2 m resolution, showcases commendable 92 % accuracy F1-score 91 %, advantages terms efficiency have been verified through large number experiments. process explores how use fundamental models accelerate remote sensing information extraction process, crucial current trajectory deep learning sensing. relevant code available https://github.com/Github-YRQ/LPL.

Language: Английский

Citations

8
FieldSeg: A scalable agricultural field extraction framework based on the Segment Anything Model and 10-m Sentinel-2 imagery DOI Creative Commons
Lucas Borges Ferreira, Vitor S. Martins, Uilson Ricardo Venâncio Aires

et al.

Computers and Electronics in Agriculture, Journal Year: 2025, Volume and Issue: 232, P. 110086 - 110086

Published: Feb. 15, 2025

Language: Английский

Citations

0
A large-scale ultra-high-resolution segmentation dataset augmentation framework for photovoltaic panels in photovoltaic power plants based on priori knowledge DOI
Ruiqing Yang, Guojin He, Ranyu Yin

et al.

Applied Energy, Journal Year: 2025, Volume and Issue: 390, P. 125879 - 125879

Published: April 10, 2025

Language: Английский

Citations

0
Space-scale exploration of the poor reliability of deep learning models: the case of the remote sensing of rooftop photovoltaic systems DOI Creative Commons
Gabriel Kasmi, Laurent Dubus, Yves‐Marie Saint‐Drenan

et al.

Environmental Data Science, Journal Year: 2025, Volume and Issue: 4

Published: Jan. 1, 2025

Abstract Photovoltaic (PV) energy grows rapidly and is crucial for the decarbonization of electric systems. However, centralized registries recording technical characteristics rooftop PV systems are often missing, making it difficult to monitor this growth accurately. The lack monitoring could threaten integration into grid. To avoid situation, remote sensing using deep learning has emerged as a promising solution. existing techniques not reliable enough be used by public authorities or transmission system operators (TSOs) construct up-to-date statistics on fleet. reliability comes from models being sensitive distribution shifts. This work comprehensively evaluates shifts’ effects classification accuracy trained detect panels overhead imagery. We benchmark isolate sources shifts introduce novel methodology that leverages explainable artificial intelligence (XAI) decomposition input image model’s decision regarding scales understand how affect models. Finally, based our analysis, we data augmentation technique designed improve robustness classifiers under varying acquisition conditions. Our proposed approach outperforms competing methods can close gap with more demanding unsupervised domain adaptation methods. discuss practical recommendations mapping imagery

Language: Английский

Citations

0
PV Segmenter: A frequency-guided edge-aware network for distributed photovoltaic segmentation in remote sensing imagery DOI
Siyuan Wang, Zhenfeng Shao, Dongyang Hou

et al.

Applied Energy, Journal Year: 2025, Volume and Issue: 393, P. 126137 - 126137

Published: May 20, 2025

Language: Английский

Citations

0
PVF-10: A high-resolution unmanned aerial vehicle thermal infrared image dataset for fine-grained photovoltaic fault classification DOI
Bo Wang, Qi Chen, Mengmeng Wang

et al.

Applied Energy, Journal Year: 2024, Volume and Issue: 376, P. 124187 - 124187

Published: Aug. 16, 2024

Language: Английский

Citations

3
Remote Sensing-Based Estimation of Rooftop Photovoltaic Power Production Using Physical Conversion Models and Weather Data DOI Open Access
Gabriel Kasmi, Augustin Touron, Philippe Blanc

et al.

Published: July 15, 2024

The global photovoltaic (PV) installed capacity, vital for the electric sector decarbonation, has reached 1,552.3 GWp in 2023. In France, capacity stood April 2024 at 19.9 GWp. growth of PV over a year was nearly 32% worldwide and 15.7% France. However, integrating electricity into grids is hindered by poor knowledge rooftop systems, constituting 20% France's lack measurements production stemming from these systems. This problem power referred to as observability. Using ground truth individual available an unprecedented temporal spatial scale, we show that estimating system combining solar irradiance temperature data, characteristics inferred remote sensing methods irradiation-to-electric conversion model provides accurate estimations production. Our study shows can improve observability, thus its integration grid, using little information on simple weather data.

Language: Английский

Citations

1
High Resolution (30 m) Burned Area Product Improves the Ability for Carbon Emission Estimation in Africa DOI Creative Commons
Baoye Qi, Zhaoming Zhang, Tengfei Long

et al.

Earth s Future, Journal Year: 2024, Volume and Issue: 12(10)

Published: Oct. 1, 2024

Abstract Fire significantly contributes to greenhouse gas emissions. The current global burned area (BA) products mainly have coarse native spatial resolution, which leads underestimation of BA and carbon emissions from biomass burning. Performances in Africa GABAM (30 m), MCD64A1 (500 GFED4s (0.25°), FireCCI51 (250 GFED5 (0.25°) were compared. From 2014 2020, detected the most BA, 1.58 times more than during same period. 0.09 Mkm 2 on average. 2016, an average 2.99 Africa, was 1.03 GFED4s. 2021, African derived 2.89 , 1.22 MCD64A1. increase will inevitably lead estimation Based GFED framework, we estimated vegetation burning 2021 be 1113.25 Tg, is higher GFED4s' time This shows that use high‐resolution m) estimate can effectively avoid overall fire

Language: Английский

Citations

1
Developing a Universal Spectral Index for Solar Photovoltaic Panels: A Methodology for Spatial Information Extraction from Satellite Remote Sensing Imagery DOI

Shuang He,

Qingjiu Tian, Jia Tian

et al.

Published: Jan. 1, 2024

Language: Английский

Citations

0
Data Augmentation with Generative Adversarial Network for Solar Panel Segmentation from Remote Sensing Images DOI Creative Commons
Justinas Lekavičius, Valentas Gružauskas

Energies, Journal Year: 2024, Volume and Issue: 17(13), P. 3204 - 3204

Published: June 29, 2024

With the popularity of solar energy in electricity market, demand rises for data such as precise locations panels efficient planning and management. However, these are not easily accessible; information sometimes does exist. Furthermore, existing datasets training semantic segmentation models photovoltaic (PV) installations limited, their annotation is time-consuming labor-intensive. Therefore, additional remote sensing (RS) creation, pix2pix generative adversarial network (GAN) used, enriching original resampled varying ground sampling distances (GSDs) without compromising integrity. Experiments with DeepLabV3 model, ResNet-50 backbone, GAN architecture were conducted to discover advantage using GAN-based augmentations a more accurate RS imagery model. The result fine-tuned panel trained transfer learning an optimal amount—60% GAN-generated data. findings demonstrate benefits images data, addressing issue limited datasets, increasing IoU F1 metrics by 2% 1.46%, respectively, compared classic augmentations.

Language: Английский

Citations

0