A review of remote sensing image segmentation by deep learning methods

International Journal of Digital Earth, Journal Year: 2024, Volume and Issue: 17(1)

Published: March 18, 2024

Remote sensing (RS) images enable high-resolution information collection from complex ground objects and are increasingly utilized in the earth observation research. Recently, RS technologies continuously enhanced by various characterized platforms sensors. Simultaneously, artificial intelligence vision algorithms also developing vigorously playing a significant role image analysis. In particular, aiming to divide into different elements with specific semantic labels, segmentation could realize visual acquisition interpretation. As one of pioneering methods advantages deep feature extraction ability, learning (DL) have been exploited proved be highly beneficial for precise recent years. this paper, comprehensive review is performed on remote survey systems kinds specially designed architectures. Meanwhile, DL-based applied four domains illustrated, including geography, precision agriculture, hydrology, environmental protection issues. end, existing challenges promising research directions discussed. It envisioned that able provide technical reference, deployment successful exploitation DL empowered approaches.

Language: Английский

---

A robust large-scale surface water mapping framework with high spatiotemporal resolution based on the fusion of multi-source remote sensing data

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2023, Volume and Issue: 118, P. 103288 - 103288

Published: April 1, 2023

Large-scale and dynamic surface water mapping is crucial for understanding the impact of global climate change human activities on distribution resources. Remote sensing imagery has become primary data source due to its high spatiotemporal resolution wide coverage. However, reliability current products during flood seasons limited influence clouds optical remote images. Moreover, annual seasonal cannot capture intra-month variations bodies. To address these challenges, we proposed a framework Google Earth Engine that combines multi-source data. Our can generate 10 m spatial maps at 15-day time step. We classified bodies using Sentinel-2 images classification tree algorithm, then used Sentinel-1 compensate cloudy missing areas in images, resulting seamless cloud-unaffected maps. evaluated effectiveness our six floodplains around world, experimental results demonstrate generated by outperform existing public datasets great potential hydrological applications. details dynamics with higher temporal free from cloud influence, which necessary resources management, monitoring, disaster response.

Language: Английский

---

A multiscale feature fusion enhanced CNN with the multiscale channel attention mechanism for efficient landslide detection (MS2LandsNet) using medium-resolution remote sensing data

International Journal of Digital Earth, Journal Year: 2024, Volume and Issue: 17(1)

Published: Jan. 4, 2024

Deep learning (DL) models have been widely used for remote sensing-based landslide mapping due to their impressive capabilities automatic information extraction. However, the large volumes of parameters and calculations compromised efficiency DL in extracting landslides from a set RS images. Lightweight convolutional neural networks (CNNs) exhibit promising feature representation abilities with fewer parameters. This study aims introduce new lightweight CNN called MS2LandsNet, designed detect both high accuracy. The MS2LandsNet consists three down-sampling stages embedded multi-scale fusion (MFF), aiming decrease while aggregating contextual features. Additionally, we incorporate channel attention (MSCA) into MFF improve performance. According experimental results on landslip datasets, obtains highest F1 score 85.90% IoU 75.28%. Notably, accomplishes resuts fewest fastest inference speed, outperforming seven classical semantic segmentation CNNs. proposed model holds potential application cloud computing platform larger-scale tasks future work.

Language: Английский

---

Monitoring land changes at an open mine site using remote sensing and multi-spectral indices

Heliyon, Journal Year: 2025, Volume and Issue: 11(2), P. e41845 - e41845

Published: Jan. 1, 2025

Language: Английский

---

Artificial Intelligence in Hydrology: Advancements in Soil, Water Resource Management, and Sustainable Development

Sustainability, Journal Year: 2025, Volume and Issue: 17(5), P. 2250 - 2250

Published: March 5, 2025

Hydrology relates to many complex challenges due climate variability, limited resources, and especially, increased demands on sustainable management of water soil. Conventional approaches often cannot respond the integrated complexity continuous change inherent in system; hence, researchers have explored advanced data-driven solutions. This review paper revisits how artificial intelligence (AI) is dramatically changing most important facets hydrological research, including soil land surface modeling, streamflow, groundwater forecasting, quality assessment, remote sensing applications resources. In AI techniques could further enhance accuracy texture analysis, moisture estimation, erosion prediction for better management. Advanced models also be used as a tool forecast streamflow levels, therefore providing valuable lead times flood preparedness resource planning transboundary basins. quality, AI-driven methods improve contamination risk enable detection anomalies, track pollutants assist treatment processes regulatory practices. combined with open new perspectives monitoring resources at spatial scale, from forecasting storage variations. paper’s synthesis emphasizes AI’s immense potential hydrology; it covers latest advances future prospects field ensure

Language: Английский

---

From perception to action: Participatory water risk assessment in Nagaon District of Assam, India

Environmental Science & Policy, Journal Year: 2024, Volume and Issue: 160, P. 103862 - 103862

Published: Aug. 13, 2024

Language: Английский

---

Analysis of Formal Concepts for Verification of Pests and Diseases of Crops Using Machine Learning Methods

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 19902 - 19910

Published: Jan. 1, 2024

This article is devoted to a set of important areas research: the analysis formal representations and verification pests pathogens affecting crops using spectral brightness coefficients (SBR) for period from 2021 2023. The database contains about 10,000 records covering growing season, types diseases pests, as well their growth phases in real coordinate system. work uses machine learning techniques including logistic regression, extreme gradient boosting (XGBoost), Vanilla convolutional neural network (CNN) analyze data classify presence satellite images. main goal optimize improve quality agricultural productivity through early detection accurate classification sector. results study can be applied development innovative systems that will increase yields, reduce cost pest disease control, production processes. conclusions this used both scientific practical recommendations enterprises organizations new technologies programs automating use promises significant breakthroughs sector, helping efficiency, sustainability, crop production.

Language: Английский

---

WaterHRNet: A multibranch hierarchical attentive network for water body extraction with remote sensing images

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2022, Volume and Issue: 115, P. 103103 - 103103

Published: Nov. 11, 2022

Water is a kind of vital natural resource, which acts as the lifeblood ecosystem and energy source for living production activities humans. Regularly mapping conditions water resources taking effective measures to prevent them from pollutions shortages are very important necessary maintain sustainability ecosystem. As preliminary step image-based resource analysis, complete recognition accurate extraction bodies prerequisites in many applications. Nevertheless, due issues topology diversities, appearance variabilities, land cover interferences, there still large gap achieve human-level interpretation quality. This paper presents hierarchical attentive high-resolution network, abbreviated WaterHRNet, extracting remote sensing imagery. First, by building multibranch feature extractor integrated with global semantics aggregation, WaterHRNet behaves laudably supply high-quality, strong-semantic representations. Furthermore, inlaying an attention scheme comprehensive exploitation both spatial channel significances, forced strengthen semantic-determinate, task-aware encodings. In addition, designing processing principle progressive enhancement category-attentive semantics, performs effectively export semantic-discriminative, target-oriented representations precise body segmentation. The elaborately verified quantitatively qualitatively on three datasets. Evaluation results show that achieves average precision 98.44%, recall 97.84%, IoU 96.35%, F1-score 98.14%. Comparative analyses also demonstrate superior performance excellent feasibility segmenting bodies.

Language: Английский

---

Mapping intertidal topographic changes in a highly turbid estuary using dense Sentinel-2 time series with deep learning

ISPRS Journal of Photogrammetry and Remote Sensing, Journal Year: 2023, Volume and Issue: 205, P. 1 - 16

Published: Oct. 1, 2023

Intertidal mudflats are an important component of the coastal geomorphological system at interface between ocean and land. Accurate up-to-date mapping intertidal topography high spatial resolution, tracking its changes over time, essential for habitat protection, sustainable management vulnerability analysis. Compared with ground-based or airborne terrain mapping, satellite-based waterline method is more cost-effective constructing large-scale topography. However, accuracy affected by extraction waterlines calibration height. The blurred boundary turbid water in tide-dominated estuary brings enormous challenges accurate extraction, errors estuarine level simulations prevent direct heights. To address these issues, this paper developed a novel deep learning using parallel self-attention mechanism boundary-focused hybrid loss to extract accurately from dense Sentinel-2 time series. UAV photogrammetric surveys were employed calibrate heights rather than simulated levels, such that error propagation constrained effectively. Annual topographic maps Yangtze China generated 2020 2022 optimized method. Experimental results demonstrate proposed could achieve excellent performance land segmentation time-varying tidal environments, better generalization capability compared benchmark U-Net, U-Net++ U-Net+++ models. comparison observations resulted RMSE 13 cm, indicating effectiveness monitoring morphological mudflats. successfully identified hotspots mudflat erosion deposition. Specifically, connected predominantly experienced deposition 10–20 cm two-year period, whereas offshore sandbars exhibited instability significant 20–60 during same period. These serve as valuable datasets providing scientific baseline information support decisions.

Language: Английский

---

Synergistic Use of Earth Observation Driven Techniques to Support the Implementation of Water Framework Directive in Europe: A Review

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(8), P. 1983 - 1983

Published: April 9, 2023

The development of a sustainable water quality monitoring system at national scale remains big challenge until today, acting as hindrance for the efficient implementation Water Framework Directive (WFD). This work provides valuable insights into current state-of-the-art Earth Observation (EO) tools and services, proposing synergistic use innovative remote sensing technologies, in situ sensors, databases, with ultimate goal to support European Member States effective WFD implementation. proposed approach is based on recent research scientific analysis six-year period (2017–2022) after reviewing 71 peer-reviewed articles international journals coupled results 11 European-founded projects related EO WFD. Special focus placed data sources (spaceborne, situ, etc.), sensors use, observed Quality Elements well computer science techniques (machine/deep learning, artificial intelligence, etc.). combination different technologies can offer, among other things, low-cost monitoring, an increase monitored per body, minimization percentage bodies unknown ecological status.

Language: Английский

---