Page run by a Geospatial Engineer. Spatial Data, GIS & Remote Sensing | Exploring the intersection of ML & AI with spatial Tech ๐ #GIS #RemoteSensing #AI #ML
Joined August 2024
- Tweets1K
- Following145
- Followers297
- Likes4.7K
Geo retweeted
๐๐ฎ๐ข๐ฅ๐๐ข๐ง๐ ๐๐ฅ๐๐ฌ๐ฌ๐ข๐๐ข๐๐๐ญ๐ข๐จ๐ง ๐ฐ๐ข๐ญ๐ก ๐๐๐จ๐๐ โ ๐๐จ๐จ๐ ๐ฅ๐ ๐๐ฅ๐ฉ๐ก๐๐๐๐ซ๐ญ๐ก ๐ข๐ง ๐๐ฒ๐ญ๐ก๐จ๐ง
Here comes my first GeoAI tutorial from 2025 - also warch out to my recent full course!
In the tutorial, we start by downloading the AlphaEarth data directly in Python, without any third-party tools, and then combine it with OpenStreetMap building footprints.
Quick recap: Google AlphaEarth is Googleโs brand-new GeoAI Earth observation initiative that combines petabytes of satellite imagery and remote sensing to present large-scale, global, 10x10m-resolution multispectral (64 bands for starters) data.
Leveraging this kind of data, due to its small-scale spatial resolution and high feature variance, it's probably going to be a game-changer for data-driven urban planning. It allows data scientists and city planners alike to characterize urban-scale areas all around the world with high precision and harness a rich set of features for planning, predictive modeling, and many other applications - especially if combined with the right vector data.
๐ ๐๐๐ฅ๐ค-๐ญ๐ก๐ซ๐จ๐ฎ๐ ๐ก ๐จ๐ง ๐๐จ๐ฎ๐ญ๐ฎ๐๐: zurl.co/PV6kC
๐ ๐๐จ๐๐ ๐จ๐ง ๐๐ฎ๐๐ฌ๐ญ๐๐๐ค: zurl.co/fj64n
๐๐ก๐ ๐๐๐ฐ ๐๐๐ข๐๐ง๐๐ ๐จ๐ ๐๐๐ฉ๐ฌ - thenewscienceofmaps.com/
๐๐จ๐จ๐ ๐ฅ๐ ๐๐ฅ๐ฉ๐ก๐๐๐๐ซ๐ญ๐ก: zurl.co/9hNwD
#30DayMapChallenge #GeoAI #Python #GeospatialData #DataVisualization #MachineLearning #GIS #Mapping #OpenStreetMap #EarthEngine #AI #BuildingClassification #SpatialAnalysis #UrbanPlanning
Geo retweeted
Have you seen Google Flood Hub yet?
Web explorer: lnkd.in/gQQ73UgV
๐๐ก๐ ๐๐๐ฐ ๐๐๐ข๐๐ง๐๐ ๐จ๐ ๐๐๐ฉ๐ฌ - thenewscienceofmaps.com/
Thanks for the infographics and the news to @Saurav lnkd.in/d2qydGyq
Geo retweeted
Today Iโve used Gemini in a Google Colab project and used Copilot for GitHub with Visual Studio Code, to make some small changes to complex projects created by other people. Iโm giving myself a Level 1 Vibe Coders Badge (Created for me by Grok) @SpatialTidbits
Geo retweeted
Language Models Guide: Transformer Model Introduction
tidbitsgeospatial.wordpress.โฆ
Geo retweeted
One cool use for Atlas is creating "bullet time"-style effects using only a few cameras.
In this case, I was able to use 3 iPhones + my finely honed juggling skillz to synthesize a smooth camera trajectory between each pair of cameras.
Also I layered in an AI-generated beats because that's just the world we live in now. Deal with it.
Geo retweeted
atlas can reconstruct berkeley too... campanile flyaround from 10 ground level images
cc @debfx
Geo retweeted
our new model Atlas also happens to be a capable text-to-image generator, providing it with a strong foundation of โworld knowledgeโ...
this generalizes across both single and multi view domains, meaning you can step into even highly stylized scenes with precise 3D control
Geo retweeted
I very highly recommend reading this post and its follow up!
It's a short, simple explanation of how GPUs work and what kernels are
Great job @MainzOnX !
Geo retweeted
Read more in the technical report: arxiv.org/abs/2608.26088
More in our blog: goo.gle/4qIAlwM
Geo retweeted
๐ Intelligent geospatial data selection: Translates natural-language prompts into strict geographic constraints, conducting grounded signal discovery across established repositories like Data Commons and Google Earth Engine, alongside live open-web discovery of government and academic portals.
๐ Multimodal dataset curation: Fuses assembled covariates with pre-trained geospatial foundation model embeddings, including Population Dynamics Foundation Models (PDFM) and AlphaEarth, while implementing an automated "Feature Gate" to prevent data leakage.
๐ Automated model building and prediction: Searches multiple model families โ such as regularized linear models, gradient-boosted decision trees, and multi-layer perceptrons โ with built-in overfitting guards and self-correction loops.
Geo retweeted
Today we are introducing the planetary prediction engine (PPE) ๐, an experimental research capability within our broader Google Earth AI initiative to turn planetary information into actionable insights directly from natural-language queries.
How the planetary prediction engine works๐
Geo retweeted
One of the hardest things I did over the past few months was to build a course to teach agentic coding techniques for geospatial problems. After endless hours of testing, refining, and writing, the full course is now launched - free and open to everyoneโก๏ธcourses.spatialthoughts.com/โฆ
Geo retweeted
Geospatial engineering:
A multidisciplinary profession integrating expertise from Geographic Information Systems (GIS), surveying, remote sensing, cartography, spatial data science, digital mapping, spatial data infrastructure (SDI), land administration, environmental monitoring
Geo retweeted
It's great honour to present a couple of projects and my journey on building regional LMs for African languages. Thank you @DeepIndaba.
You can find the slides on my website dadelani.github.io/talks.htmโฆ
Happy to answer further questions via email or at Mila booth today (Aug 4)
"Many low-resource African languages are still not prioritised in how these technologies are built." ๐
David Adelani @davlanade from @Mila_Quebec took #DLI2026 through lessons from the trenches: AfroXLMR, AfriTeVa V2, and AfriqueLLM, a large language model spanning 20 African languages. The call: build African-centric models for African languages.
#DLI2026 #Indaba2026 #รmแปWa #sovereignintelligence
Geo retweeted
We're hiring a Research Engineer at @Arsenal โฝ๐ดโช to work directly with our Men's First Team!
We're building state-of-the-art AI models for the football domain. This role will focus on building the application layer for our research to advance coaching and analysis workflows.
Geo retweeted
weird to think ML research once meant spending time learning about matplotlib Figure vs Axes
