HARVEST in Action
Takeoff
The DJI Mavic Air 2S leaves its dock (note: the dock will be constructed when sufficient funds are acquired) to begin its scheduled scan. The camera begins broadcasting to the cloud, and on the farmer’s side of the application, the live feed of the current field data flickers to life.
Object Detection
We annotated dozens of pictures by hand to train HARVEST to recognize and categorize areas of cropland that contain either high, low, or no maturity crops. By identifying which crops display stunted growth patterns, we’re able to help farmers direct their watering efforts to strategic areas of their crop fields. The farmer receives updates about where to water their field from the application’s live data.
Landing
After completing its scan, the drone returns to its dock (or just the ground, for now) using geotagging and RTH (return-to-home) technology. It stays in its dock until the next scanning cycle, where it will charge protected from the elements. This fully automated design allows for a hands-off experience for the farmer.
Response
Now equipped with all the necessary information, the farmer can adjust their watering cycle to cater to the identified problem areas. They don’t need to change their irrigation technology at all — all they need to do is water the plants as indicated by our algorithm. Tackling critical regions this way, rather than flooding the entire field, significantly reduces water waste, contributing towards a more sustainable future.