#CopterSpotterVision uses computer vision to programmatically identify the helicopters of DC.

Since the end of 2020 we have been logging and annotating submitted helicopter photos using Roboflow. The DC based tech startup makes collaborative annotation of computer vision data seta breeze. But the real game changer for me was their Roboflow Train feature which uses Google's AutoML technology to let even non-coders quickly and easily roll out computer vision implementations.

The benefit for #CopterSpotter users is we can automatically process unknown spots through our model to make an educated guess at what type of helicopter was photographed. On spots that utilize this inference you will see a percentage confidence score afterwards, hopefully this will give users a window into what's happening programmatically and of course we hope to see this confidence score grow as we enhance the model. @HelicoptersofDC will also reply with ID results to any picture regardless of location and 🚁 normally required for #CopterSpotter logging - time to submit your most mysterious helicopter photos!

If you want to learn more our dataset is public on Roboflow's Universe sandbox, all you have to do is create a free account to start adapting it to your own needs:

Release Notes:

August 5th 1.2 

Now 12,000+ images, added many photos of previously under-represented classes R66, VH92A, S76

Added "CH47" class, added "bird" class, added "plane" class, added null examples

Rotation: Between -10° and +10°

May 6th 1.1 - Doubled images to 4061, specifically adding previously under-represented classes such as R66, V22 and A139 - added image augmentations:

Flip: Horizontal

Rotation: Between -2° and +2°

Saturation: Between -50% and +50%

Exposure: Between -25% and +25%

1.0 - the beta test version comprised of over 2000 images