In recent years machine learning and artificial intelligence has revolutionized the way we analyze the world from above. The ability to automatically detect and identify objects in satellite images has unlocked a myriad of possibilities, including monitoring deforestation, tracking wildlife, and enhancing disaster response.
Some recent examples of machine learning being used to search aerial imagery include OneSoil (which uses AI to detect the types of crops being grown), Земляна проказа (identifying illegal amber mines) and Curio Canopy (identifying tree canopy cover in European cities).
Robin Wilson has released an impressive Aerial Image Search Demo which allows you to explore an AI image search of aerial imagery for yourself. Using the map you can search an image of Southampton for any object that you want. For example in the screenshot at the top of this post you can see the results of my search for 'boats'.
The map says that under the hood it uses "the SkyCLIP vector embedding model, and Pinecone vector DB". If you are interested in developing your own AI satellite image search there are links to both models. Meanwhile you can have fun searching Southampton for shipping containers, churches, crops, car parks, swimming pools - or anything else that you can think of.
Unfortunately my own searches of the map reveal that Southampton may be harboring a large number of illegal amber mines and nuclear weapon silos.
Via: Weekly OSM
No comments:
Post a Comment