Classifying satellite imagery - Made with TensorFlow.js
HTML-код
- Опубликовано: 10 сен 2024
- Meet Sean McGee, a software developer at Esri UK who helps customers solve real-world problems with GIS (Geographical Information Systems) and machine learning. Recently he has been using TensorFlow.js with satellite imagery to recognize objects across geographical regions client side in the browser.
Try it for yourself:
Github Repository → goo.gle/3AuMxGy
GIS Satellite Imagery by Sean McGee → goo.gle/3lLZk2y
Want to be on the show? Use #MadeWithTFJS to share your own creations on social media and we may feature you in our next show.
Catch more #MadeWithTFJS interviews → goo.gle/made-wi...
Subscribe to the TensorFlow channel → goo.gle/Tensor...
product: TensorFlow - TensorFlow JS, TensorFlow - General; fullname: Jason Mayes;
Very interesting. I worked with classified sat imagery in the late 70s and early 80s. I was a photo interpreter who worked in a deep hole somewhere is eastern Nebraska. This is amazing technology. Finding potholes sooner would surely reduce road maintenance costs over the long term. I would think traffic control would also be a user of this technology.
Thank you for sharing! Indeed I think this sort of thing could bring many efficiencies in this industry. If you know folk in this space please do share the video with them, and of course always happy to speak to folk if they want to connect with me on LinkedIn if they have questions about TensorFlowJS!
USING SATELITES TO FIND POTHOLES!!!? hahahaha!!! the government knows where the potholes are, they just don't want to fix them. don't be naive
@@davidanalyst671 maybe in America, in the UK is actually pretty decent and Germany has roads so good there are no speed limits.
Thanks for sharing on the integration of GIS and ML
Lovely one there Sir @Jason
Cool project. Real time filters on real time stream would be future.
Hey sir is there a detailed video to run this full thing as i am new getting confused. Please let me know.
Hi, is it possible to calculate the number of signals for a specific radius in US and Canada and what's the model I need to use. Is it possible to achieve with Google maps and not writing a single JS code
Sir, this project runs on images. Are there similar projects which runs on videos? Kindly reply.
Yes many of our models run in real time so you can just call many times per second as video is just 30 images per second. 🎉 This however is the first one I've seen applied to GIS but no reason why you could not make a similar thing for video
It should be absolutely possible! Not something I've personally tried yet though
From a concept pov, that would be similar to real-time traffic cam detection
Interesting, is applicable counting palm coconut tree?
You could certainly train to recognize such trees with enough training data.
Which type of Satelite images are you using?
Looks cool, but flutter plugins and Firebase Ml custom are really slow in development.