Taylor Geospatial Institute’s Post

🤔 Exploring synthetic data for computer vision? Don't miss this 📆 webinar on July 25th at 2 pm EDT! Discover how #syntheticdata fills the data gap and helps #engineers draw more accurate, reliable results from #computervision models - all while saving money💲💲! 💥 Register now: https://buff.ly/3Wtso0V 💥 In this webinar, Rendered.ai's COO and Head of Product, Chris Andrews, will reveal: ✅ The advantages of using synthetic data to train #geospatial #artificialintelligence / #machinelearning systems ✅ Real examples of how #GEOINT professionals are using synthetic data today ✅ Expert tips to advance your use of synthetic data, including a demo of using AI tools to clean imagery backgrounds for enhanced synthetic data generation Save your seat before it's too late! Space Capital Union Labs Ventures Marlinspike Congruent Ventures Tectonic Ventures IQT (In-Q-Tel) #NVIDIAInception AI Partnerships Corp. The Open Geospatial Consortium (OGC) Air & Space Forces Association IEEE Computer Society Computer Vision Foundation Geospatial World Forum (GWF) United States Geospatial Intelligence Foundation (USGIF)

🚀 Are you exploring synthetic data to help power geospatial computer vision models? Join Rendered.ai's insightful webinar hosted by COO and Head of Product, Chris Andrews, on Thursday July 25th at 2:00 p.m. EDT. Discover: 🌟 The advantages of using synthetic data to train AI/ML systems. 🌟 Real-world examples of how GEOINT professionals leverage synthetic data today. 🌟 Tips for generating higher-quality synthetic data, including a demo on using AI tools to clean imagery backgrounds in the Rendered.ai PaaS. 🎟️ SAVE YOUR SEAT: register now to gain expert insights and ask your most pressing questions about synthetic data! #aipartnershipscorp #artificialintelligence #machinelearning #webinar

"Exciting news in the world of computer vision and satellite imagery! The CloudTracks dataset has been introduced, containing over 12,000 ship track annotations in 3,560 satellite images. This dataset has enabled the development of more accurate and efficient models for localizing and counting ship tracks, with significant improvements over previous state-of-the-art methods. The release of this dataset is a major step forward in advancing machine learning approaches for detecting elongated and overlapping features in satellite images. Check out the details and access the dataset at https://lnkd.in/dvkzg3yq. #ComputerVision #SatelliteImagery #MachineLearning #CloudTracks"

Foundation models that integrate multiple modalities of data are becoming increasingly common for a variety of domains. Xin Guo et al. proposed a model called SkySense for earth observation imagery, including temporal sequences of optical and SAR. They included a number of enhancements to typical multimodal models, including Multi-Granularity Contrastive Learning to learn representations for different modalities and spatial resolutions and Geo-Context Prototype Learning for learning region-aware prototypes. With these novelties, SkySense was able to outperform a number of alternative models on 16 different datasets. The authors included a small ablation study on one dataset, quantifying the benefits of each component of their model. I'd love to see a more extensive ablation study across multiple datasets to confirm what the most important components of this model are. https://lnkd.in/eEksNwBF For more info on how you can bring the latest research models into action on your data, sign up for my Computer Vision Insights newsletter: https://lnkd.in/g9bSuQDP #RemoteSensing #EarthObservation #MachineLearning #DeepLearning #ComputerVision

🌍 I am excited to announce the release of our research project: “MMEarth: Exploring Multi-Modal Pretext Tasks For Geospatial Representation Learning.” 🚀 It is the outcome of my research visit at the Pioneer Centre for AI in Copenhagen! Our new dataset “MMEarth” is now online and ready for exploration. It’s a diverse multi-modal pre-training dataset for remote sensing data at global scale. Check out our preprint and the project page for details on the dataset and our proposed model: https://lnkd.in/daQqsyr7 Here are some key highlights: • Global Coverage: MMEarth spans data from 1.2 million locations worldwide, sampled uniformly across 14 biomes, akin to ImageNet-1K in optical image count. • Multi-Modal Data: At each location, we’ve collected data from 12 aligned modalities • Taster Datasets: We provide 2 taster datasets to facilitate further research in multi-modal representation learning. • Experiments: Our initial experiments focus on learning general-purpose representations for Sentinel-2 images by leveraging other modalities as pretext tasks. Big thanks to Nico Lang, Ankit K., Stefan Oehmcke, Christian Igel, and Serge Belongie for their invaluable contributions and guidance throughout the project. Stay tuned for updates, and don’t hesitate to reach out if you have any questions or want to discuss MMEarth! 🛰️🌐

Satellogic open-source release: A large dataset of high-resolution imagery for AI model training [https://lnkd.in/d9hBX5AW] Satellogic announced the release of a large open dataset of high-resolution imagery curated from its catalog to support the training of AI models. The dataset contains around 3 million 384m by 384m Satellogic images of unique locations (6 million images, including location revisits) from around the world – totaling 900 Gigapixels spanning different land-use types, objects, geographies, and seasons. Satellogic data is released under a Creative Commons CC-BY 4.0 license, allowing for commercial use of the data with attribution. The full dataset can now be accessed on Hugging Face as part of EarthView (https://lnkd.in/dc2q4Xxv) Source: Akis Karagiannis (https://lnkd.in/dqFC5gPu)

😢 If you missed the "Powering Geospatial Intelligence with Synthetic Data" webinar last week, don't worry! It's now available to watch on demand👏: https://buff.ly/3YlpcG4 Gain valuable insights from Rendered.ai experts on: ✅ The advantages of training #artificialintelligence and #machinelearning systems with synthetic data ✅ Real-world examples of how #geospatial intelligence professionals are using synthetic data in #computervision today ✅ How to use AI tools to clean imagery backgrounds more easily for #syntheticdata generation in the Rendered.ai platform ⬇️ Comment and let us know what you'd like us to cover in future webinars! Space Capital Union Labs Ventures Marlinspike Congruent Ventures Tectonic Ventures IQT (In-Q-Tel) #NVIDIAInception AI Partnerships Corp. The Open Geospatial Consortium (OGC) United States Geospatial Intelligence Foundation (USGIF) Geospatial World Forum (GWF)

i am happy to tell you that we’ve been working on improving road extraction from satellite imagery using a Residual U-Net architecture. By integrating data augmentation techniques, our model shows enhanced robustness and accuracy in segmenting roads.📈 This approach leverages deep learning with residual blocks, batch normalization, and customized loss functions to deliver precise results. Our experimental findings demonstrate the model's capability in accurately delineating road networks. For those interested in advancements in remote sensing and computer vision, our research offers an innovative solution worth exploring. to read our research article visit my research publication portion on my profile. 🎆 #DeepLearning #ComputerVision #RemoteSensing #UNet #DataAugmentation #AI #MachineLearning #SatelliteImagery #Innovation #Research

Geospatial artificial intelligence is the application of AI, geospatial data, science, and technology to accelerate real-world understanding of business opportunities, environmental impacts, and operational risks. Through automated data generation, algorithms and GIS tools, CITY@lums is optimizing how cities function. How can we use GeoAI to design resilient cities? 💡 Turn urban data into information, with models that adapt even as data evolves.  💡 Improve data quality, consistency, and accuracy 💡 Streamline manual data generation workflows by using the power of automation to increase efficiency and reduce costs. 💡 Accelerate situational awareness 💡 Monitor and analyze climate and environmental events, assets, and entities from sensors and sources to enable quicker response times and proactive decisions. 💡 Bring location intelligence to decision-making 💡 Make data-driven decisions with real-world awareness. 💡 Improve outcomes with insight from spatial patterns and accurate predictions.

Upcoming #webinar alert: Redefining Image Analysis with ENVI 6.0, IDL 9.0, and the ENVI Ecosystem. As the way people interact with imagery has changed, so has NV5 Geospatial! Join this webinar to learn how their new product releases will transform the way that you, and your organization, work with imagery. Attend this webinar to learn: * New image and #SAR processing #workflows that make science approachable * How users can improve productivity with new workflows and developer tools * A new playground for data scientists powered by #IDL Notebooks * How experts and non-experts can easily collaborate to solve #geospatial problems * How you can use AI to super-charge video and image analysis There will be Q&A at the end of the webinar to answer your questions. Date: Tuesday, 30th November 2023 Time: 3:00 PM GMT (for the EMEA region) Click here to register: https://bit.ly/3uhlwIj #envi #remotesensing #geospatialtechnology #datascientists