Sentinel-1
PulseAugur coverage of Sentinel-1 — every cluster mentioning Sentinel-1 across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
AI predicts avalanche activity using snowpack simulations and satellite data
Researchers have developed a data-driven approach using a Transformer++ model to predict avalanche activity by analyzing snowpack simulations and satellite data. The model was trained on four winters of Sentinel-1 synth…
-
New benchmark Infra-Bench CLS tests foundation models on critical infrastructure classification
A new benchmark, Infra-Bench CLS, has been introduced to evaluate the effectiveness of Earth observation foundation models in classifying critical infrastructure. The benchmark comprises 18,756 images of facility-scale …
-
Geospatial foundation models show promise in tree species mapping
Researchers have developed a new method for mapping tree species in Denmark by comparing traditional spectral-temporal features with embeddings from geospatial foundation models like TESSERA and AlphaEarth. The study fo…
-
Foundation models show promise in mapping Danish tree species
Researchers have developed a method to map tree species across Denmark using satellite data and machine learning. The study compared manually engineered spectral-temporal features (STF) with embeddings from foundation m…
-
New AI model maps forest canopy height with high resolution using public satellite data
Researchers have developed SERA-H, a novel deep learning model designed for high-resolution canopy height mapping using publicly available satellite data. This model integrates a super-resolution module (EDSR) with temp…
-
Satellite data integration yields accurate building height estimates for Global South cities
Researchers have developed a method for estimating building heights at an individual footprint scale, particularly for urban areas in the Global South where detailed data like LiDAR is scarce. By integrating data from T…
-
New AI framework CREST segments polar lows in SAR imagery
Researchers have developed a novel weakly supervised semantic segmentation framework called CREST to identify polar lows in Sentinel-1 SAR imagery. This method addresses the challenge of limited pixel-level masks by gen…
-
Middle East oil flows defy war, stabilize global prices · 1 source tracked
Middle Eastern oil producers are maintaining significant crude oil exports through the Strait of Hormuz, despite ongoing conflict in the region. These covert shipments, often disguised as "dark" transfers, are helping t…
-
Deep embeddings boost tree species classification accuracy in Dutch forest inventory
A new research paper explores the use of deep embeddings from pre-trained remote sensing models to improve tree species classification in the Netherlands' National Forest Inventory. The study found that these deep embed…
-
New technique aligns satellite data for improved Antarctic sea ice labeling
Researchers have developed a new method to align multimodal satellite imagery, specifically from Sentinel-1 and MODIS platforms, to improve the labeling of Antarctic sea ice. This approach addresses the challenge of spa…
-
New CNN model enhances biomass estimation using multi-sensor data
Researchers have developed a novel convolutional neural network (CNN) model for estimating above-ground biomass (AGB) using multi-sensor data, including optical, SAR, and terrain information. This globally trained model…
-
Allen AI releases OlmoEarth embeddings for Earth observation analysis
Allen AI has released OlmoEarth embeddings, which are numerical representations of Earth observation data derived from their open-source OlmoEarth foundation models. These embeddings, available through the OlmoEarth Stu…
-
New Deep Evidential Regression method estimates forest height with uncertainty
Researchers have developed a new method called Deep Evidential Regression (DER) to estimate forest height from satellite imagery, which also quantifies predictive uncertainty. This approach is particularly useful for sp…
-
New SLED method offers scalable, cost-effective geospatial data encoding
Researchers have developed a new method called Scalable Location Encoding via Distillation (SLED) for creating efficient location encoders from geospatial data. Unlike previous methods that rely on computationally expen…
-
Deep learning models benchmarked for offshore wind infrastructure monitoring
Researchers have benchmarked various deep learning models for classifying events related to offshore wind infrastructure using Sentinel-1 satellite data. The study compared ten different model training variants, includi…
-
New open dataset OSSDD released for Sentinel-1 ship detection
Researchers have introduced OSSDD, a new open dataset designed for training neural networks to detect ships in Synthetic Aperture Radar (SAR) images. This dataset, built upon the OpenSARShip 1.0 dataset, provides 15,197…
-
PhenoStitch pipeline maps crops without task-specific training
Researchers have developed PhenoStitch, a novel pipeline for panoptic crop mapping using satellite imagery that eliminates the need for extensive task-specific training. The system first employs a frozen Segment Anythin…
-
New GEOID-Flood dataset advances AI-driven flood segmentation
Researchers have introduced GEOID-Flood, a new large-scale, multi-modal benchmark dataset designed for flood segmentation tasks. This dataset, derived from over ten years of Copernicus Emergency Management Service activ…
-
Space2Ground 2.0 fuses street-level and satellite imagery for enhanced agricultural monitoring
Researchers have developed Space2Ground 2.0, a new framework and dataset designed to improve agricultural monitoring by fusing street-level imagery with satellite data. This system processes large volumes of crowdsource…
-
TerraMind vs THOR: Architectural Differences Drive GFM Performance
A new research paper systematically compares two Geospatial Foundation Models (GFMs), TerraMind and THOR, developed under the European Space Agency's $\Phi$-lab. The study moves beyond aggregate scores to analyze archit…