AI-Powered Satellite Imagery Processing for Global Air Traffic Surveillance

LatIA, Journal Year: 2025, Volume and Issue: 3, P. 80 - 80

Published: Feb. 19, 2025

The increasing complexity of global air traffic management requires innovative surveillance solutions beyond traditional radar. This chapter explores the integration artificial intelligence (AI) and machine learning (ML) in satellite imagery processing for enhanced surveillance. proposed AI framework utilizes remote sensing, computer vision algorithms, geo-stamped aircraft data to improve real-time detection classification. It addresses limitations conventional systems, particularly areas lacking radar coverage. study outlines a three-phase approach: extracting coverage from imagery, labeling with locations, applying deep models YOLO Faster R-CNN distinguish other objects high accuracy. Experimental trials demonstrate AI-enhanced monitoring's feasibility, achieving improved high-traffic zones. system enhances situational awareness, optimizes flight planning, reduces airspace congestion, strengthens security. also aids disaster response by enabling rapid search-and-rescue missions. Challenges like adverse weather nighttime monitoring remain, requiring infrared sensors radar-based techniques. By combining big analytics, cloud computing, monitoring, offers scalable, cost-effective solution future management. Future research will refine expand predictive analytics autonomous surveillance, revolutionizing aviation safety operational intelligence.

Language: Английский

---

Segment Anything Model Can Not Segment Anything: Assessing AI Foundation Model’s Generalizability in Permafrost Mapping

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(5), P. 797 - 797

Published: Feb. 24, 2024

This paper assesses trending AI foundation models, especially emerging computer vision models and their performance in natural landscape feature segmentation. While the term model has quickly garnered interest from geospatial domain, its definition remains vague. Hence, this will first introduce defining characteristics. Built upon tremendous success achieved by Large Language Models (LLMs) as for language tasks, discusses challenges of building artificial intelligence (GeoAI) tasks. To evaluate large Meta’s Segment Anything Model (SAM), we implemented different instance segmentation pipelines that minimize changes to SAM leverage power a model. A series prompt strategies were developed test SAM’s regarding theoretical upper bound predictive accuracy, zero-shot performance, domain adaptability through fine-tuning. The analysis used two permafrost datasets, ice-wedge polygons retrogressive thaw slumps because (1) these landform features are more challenging segment than man-made due complicated formation mechanisms, diverse forms, vague boundaries; (2) presence important indicators Arctic warming climate change. results show although promising, still room improvement support AI-augmented terrain mapping. spatial generalizability finding is further validated using general dataset EuroCrops agricultural field Finally, discuss future research directions strengthen applicability domains.

Language: Английский

---

Towards Responsible Urban Geospatial AI: Insights From the White and Grey Literatures

Journal of Geovisualization and Spatial Analysis, Journal Year: 2024, Volume and Issue: 8(2)

Published: June 26, 2024

Abstract Artificial intelligence (AI) has increasingly been integrated into various domains, significantly impacting geospatial applications. Machine learning (ML) and computer vision (CV) are critical in urban decision-making. However, AI implementation faces unique challenges. Academic literature on responsible largely focuses general principles, with limited emphasis the domain. This important gap scholarly work could hinder effective integration Our study employs a multi-method approach, including systematic academic review, word frequency analysis insights from grey literature, to examine potential challenges propose strategies for (GeoAI) integration. We identify range of practices relevant complexities using planning its implementation. The review provides comprehensive actionable framework adoption domain, offering roadmap researchers practitioners. It highlights ways optimise benefits while minimising negative consequences, contributing sustainability equity.

Language: Английский

---

An ensemble framework for explainable geospatial machine learning models

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2024, Volume and Issue: 132, P. 104036 - 104036

Published: July 16, 2024

Analyzing spatially varying effects is pivotal in geographic analysis.However, accurately capturing and interpreting this variability challenging due to the increasing complexity non-linearity of geospatial data.Recent advancements integrating Geographically Weighted (GW) models with artificial intelligence (AI) methodologies offer novel approaches.However, these methods often focus on single algorithms emphasize prediction over interpretability.The recent GeoShapley method integrates machine learning (ML) Shapley values explain contribution geographical features, advancing combination ML explainable AI (XAI).Yet, it lacks exploration nonlinear interactions between features explanatory variables.Herein, an ensemble framework proposed merge local spatial weighting scheme XAI technologies bridge gap.Through tests synthetic datasets comparisons GWR, MGWR, GeoShapley, verified enhance interpretability predictive accuracy by elucidating variability.Reproducibility explored through comparison schemes various models, emphasizing necessity model reproducibility address parameter uncertainty.This works both regression classification, offering a approach understanding complex phenomena.

Language: Английский

---

Assessment of a new GeoAI foundation model for flood inundation mapping

Published: Nov. 13, 2023

Vision foundation models are a new frontier in Geospatial Artificial Intelligence (GeoAI), an interdisciplinary research area that applies and extends AI for geospatial problem solving geographic knowledge discovery, because of their potential to enable powerful image analysis by learning extracting important features from vast amounts data. This paper evaluates the performance first-of-its-kind model, IBM-NASA's Prithvi, support crucial task: flood inundation mapping. model is compared with convolutional neural network vision transformer-based architectures terms mapping accuracy flooded areas. A benchmark dataset, Sen1Floods11, used experiments, models' predictability, generalizability, transferability evaluated based on both test dataset completely unseen model. Results show good Prithvi highlighting its advantages segmenting areas previously regions. The findings also indicate improvement adopting multi-scale representation learning, developing more end-to-end pipelines high-level tasks, offering flexibility input data bands.

Language: Английский

---

Automated Text Recognition and Segmentation for Historic Map Vectorization: A Mask R-CNN and UNet Approach

Deleted Journal, Journal Year: 2024, Volume and Issue: 20(7s), P. 635 - 649

Published: May 4, 2024

Historic maps are essential for comprehending how buildings and landscapes have changed over time. For this—vectorization can be a useful method of analysis an extensive collection these maps. However, text overlaps with structural elements—often makes this process more difficult. Therefore, automated pipeline recognition, pixel-level mask creation, dataset generation, bounding box detection has been proposed. Findings shows—text segmentation, detection, recognition were demonstrated by the combination Mask Region-based Convolutional Neural Network (Mask R-CNN) UNet model achieved 99.12% all occurrences in images—which also attained accuracy 87.72% while collecting inside boxes. This end-to-end shows potential wide range future uses, especially when it comes to removal purpose making historic easier vectorize analyze—which will improve understanding historical landscapes.

Language: Английский

---

Mass-Balance-Consistent Geological Stock Accounting: A New Approach toward Sustainable Management of Mineral Resources

Environmental Science & Technology, Journal Year: 2024, Volume and Issue: 58(2), P. 971 - 990

Published: Jan. 2, 2024

Global resource extraction raises concerns about environmental pressures and the security of mineral supply. Strategies to address these depend on robust information natural endowments, suitable methods monitor model their changes over time. However, current resources reserves reporting accounting workflows are poorly suited for addressing depletion or answering questions long-term sustainable Our integrative review finds that lack a theoretical concept framework mass-balance (MB)-consistent geological stock hinders systematic industry-government data integration, governance, strategy development. We evaluate existing literature accounting, identify shortcomings monitoring mine production, outline conceptual MB-consistent system integration based material flow analysis (MFA). synthesis shows recent developments in Earth observation, geoinformation management, sustainability act as catalysts make increasingly feasible. propose first steps its implementation anticipate our perspective "resource realists" will facilitate anthropogenic systems, help secure future supply, support global transition.

Language: Английский

---

GeoAI Reproducibility and Replicability: A Computational and Spatial Perspective

Annals of the American Association of Geographers, Journal Year: 2024, Volume and Issue: 114(9), P. 2085 - 2103

Published: July 23, 2024

Language: Английский

---

Geospatial foundation models for image analysis: evaluating and enhancing NASA-IBM Prithvi’s domain adaptability

International Journal of Geographical Information Science, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 30

Published: Aug. 30, 2024

Research on geospatial foundation models (GFMs) has become a trending topic in artificial intelligence (AI) research due to their potential for achieving high generalizability and domain adaptability, reducing model training costs individual researchers. Unlike large language models, such as ChatGPT, constructing visual image analysis, particularly remote sensing, encountered significant challenges formulating diverse vision tasks into general problem framework. This paper evaluates the recently released NASA-IBM GFM Prithvi its predictive performance high-level analysis across multiple benchmark datasets. was selected because it is one of first open-source GFMs trained time-series high-resolution sensing imagery. A series experiments were designed assess Prithvi's compared other pre-trained task-specific AI analysis. New strategies, including band adaptation, multi-scale feature generation, fine-tuning techniques, are introduced integrated an pipeline enhance adaptation capability improve performance. In-depth analyses reveal strengths weaknesses, offering insights both improving developing future tasks.

Language: Английский

---

Towards AI-Assisted Mapmaking: Assessing the Capabilities of GPT-4o in Cartographic Design

ISPRS International Journal of Geo-Information, Journal Year: 2025, Volume and Issue: 14(1), P. 35 - 35

Published: Jan. 17, 2025

Cartographic design is fundamental to effective mapmaking, requiring adherence principles such as visual hierarchy, symbolization, and color theory convey spatial information accurately intuitively, while Artificial Intelligence (AI) Large Language Models (LLMs) have transformed various fields, their application in cartographic remains underexplored. This study assesses the capabilities of a multimodal advanced LLM, GPT-4o, understanding suggesting elements, focusing on established principles. Two assessments were conducted: text-to-text evaluation an image-to-text evaluation. In assessment, GPT-4o was presented with 15 queries derived from key concepts cartography, covering classification, theory, typography. Each query posed multiple times under different temperature settings evaluate consistency variability. evaluation, analyzed maps containing deliberate errors assess its ability identify issues suggest improvements. The results indicate that demonstrates general reliability text-based tasks, variability influenced by settings. model showed proficiency classification symbolization tasks but occasionally deviated theoretical expectations. hierarchy layout, performed consistently, appropriate choices. effectively identified critical flaws inappropriate schemes, poor contrast misuse shape size variables, offering actionable suggestions for improvement. However, limitations include dependency input quality challenges interpreting nuanced relationships. concludes LLMs like significant potential design, particularly involving creative exploration routine support. Their critique generate elements positions them valuable tools enhancing human expertise. Further research recommended enhance reasoning expand use variables beyond color, thereby improving applicability professional workflows.

Language: Английский

---