We’re announcing Canopy Height Maps v2 (CHMv2), an open source model for high-resolution global forest canopy mapping, developed in partnership with the World Resources Institute. CHMv2 leverages our DINOv3 Sat-L vision model, specifically optimized for satellite imagery, to deliver substantial improvements in accuracy, detail, and global consistency. 🔗 Learn more:

CHMv2 is already supporting public sector efforts in the United States, Europe, and beyond. By making these advances open source, we aim to accelerate research and inform carbon offsetting, reforestation, and land management decisions globally. 🔗 Read the paper: 🔗 Download the model:

This is bigger than forest mapping. DINOv3 + CHMv2 is essentially a planetary perception layer. Once AI can infer 3D structure from satellite pixels, Earth becomes a continuously updating digital twin. Forests today. Cities tomorrow. Infrastructure next. We’re quietly teaching machines to understand the geometry of the planet. 🦾

@WorldResources High-res global forest canopy maps can really accelerate climate change research.

@WorldResources Looks so cool, Can it find my missing cricket balls?

@WorldResources Awesome!! #AI Meta Korea B2C "Me"

@WorldResources Hi Meta. 2 days ago I payed meta business to speak with human beings after my FB was hacked ! All day long I’m waiting and nobody is chatting with me… Please give me my account back, I wanna find my thousand of tree pics back, and my friends. As a writer, novelist and poet💥😞

@WorldResources nice

The fact that independent developers now have free access to satellite-optimized vision models like DINOv3 is wild. You guys are completely removing the barrier to entry for building impactful climate and mapping tech. Time to fire up the Python environment and experiment with this 🐍🛰️

@WorldResources AI helping us understand and protect Earth is powerful.

@WorldResources dear @AIatMeta Let "AI" Artificial Intelligence do absolutely all the work for us humans, and we all live in complete prosperity, we all have a lot of things at home, lots of cars, and all we do is enjoy life and walk and travel constantly,

@WorldResources DINOv3 Sat-L is a massive unlock for geospatial AI. Open-sourcing CHMv2 is huge for global conservation.

@WorldResources wow, this is so cool.

@WorldResources so sad

This is exactly the kind of open-source contribution the planet needs right now!! Making CHMv2 freely available for reforestation and carbon science is a massive win .. the Meta AI team keeps delivering 🌱 We track how AI is reshaping sustainability work over at

@WorldResources

@WorldResources Using AI to count trees is genuinely cool and also the most wholesome thing a company burning 6GW of power has ever done. The carbon math is doing some heavy lifting here

@WorldResources

@WorldResources Any plans for improving the resolution of the maps? Currently its resolution does not allow for accurately determining the width of the three crown

@WorldResources Open sourcing CHMv2 is huge. Consistency is the hard part with global satellite models.

@sitgdev @WorldResources What's the ground truth ratio ?

@WorldResources 😊😊😊😊

@WorldResources The only way is to buy your membership to have customer service staff, but it doesn't help anything, and you can't apply for a new account. The number of victims is still expanding, but you are still bragging about your AI.

@WorldResources When will you deal with the issue of AI mass suspension of ordinary users? You don't have a customer service unit and you don't have the willingness to solve it.

@WorldResources Open source + a real deployment partner is the useful combination here. The strongest AI infra moves usually show up when research gets tied to a concrete downstream decision.

@WorldResources Satellite vision models for forest mapping > another chatbot wrapper. This is the kind of AI that actually moves the world.

@WorldResources CHMv2 uses DINOv3 Sat-L to map global forest canopy height with unprecedented detail and accuracy. Open-source. Planet-scale. 🌍

@WorldResources

@WorldResources Wait Meta 😄