Using computer vision to detect objects of interest with drone videos

[mavicair2eng]

Hey all, here's an app that can find objects of interest in a drone (Mavic Air 2) videos using a computer vision model & then plot them on a map using the flight log. This example finds solar panels, but this same thing could be used for things like search & rescue or looking at roof damage or pretty much anything else you might be looking for from the air.

The published code is here: GitHub - roboflow-ai/dji-aerial-georeferencing: Detect objects in drone videos and plot them on a map

And you can try it out with your own drone videos & flight logs here (all the processing happens in your web browser; your flight logs & video never leaves your computer): roboflow-ai/dji-aerial-georeferencing

Would love to hear what you think & if you have ideas for other ways to use/improve it!

 

[AnzacJack]

We are using software like this being used over live feed to spot sharks
Many months of “computer learning ” to differentiate between swimmers, schools of fish, whales etc

 

[mavicair2eng]

Thanks for sharing! It would be interesting to take a look at that software if you're able to share the company or any links with more information.

This project is open source so anyone is able to use or build off of it for free.