AI in Ag: Drones

At the end of May we presented at Data Tech 2019 - but the event was sold out and so we know not everyone who wanted to come was able to- so we have broken down our presentation to a few excerpts. When speaking to a AI audience without an agricultural background we try to highlight areas of continued opportunity, and ways in which commercial providers can help farmers to be successful. Creating bridges between technology and agriculture is a passion of ours. If you have a different perspective on imagery, we’d love to hear from you! Comment, message or e-mail us to share your thoughts and experiences. - Karen

Access to satellite and drone imagery is still proving its importance in agriculture - and I believe it is because of the very things that can be enabled by Artificial Intelligence.  Satellite imagery can show point in time context, such as conditions during seeding, and harvest and provide visualization of performance and yields.

Many companies are now offering satellite imagery on a daily basis for farmers to be able to analyze what their fields look like, any signs of disease, yellow of the crop (lack of nitrogen), presence of water/water distress.  This is all key, important information to farmers, but it stops short of doing automated image comparisons to highlight to the farmer where the changes are in the field, where the imagery detects that further attention needs to be paid.  


Drone imagery can enable access to parts of the field that cannot be easily scouted by foot, and can be programmed to automatically take off and fly the field, but requires image stitching/processing which can be extremely challenging when uploading data over slow internet speeds often found in rural areas.

Integration of this imagery with “action taken” would also be hugely helpful.  For example if the farmer sees there is nitrogen deficiency and goes out and does an application, being able to connect that action back to the imagery to determine if that application changes the crop healthy in the next 5-7 days is important.  

Imagery on the fields themselves is also only one application.  Imagery of the farm yard, particularly the bin sites are equally important.  Farmers store their production in bins on-site until their contract calls for delivery.  That means there can be millions of dollars sitting in a bin site awaiting delivery. Just like a grocery store has security cameras to protect their inventory, the same is needed by farmers to protect their bins.  Being able to do so, while detecting only “unrecognized” vehicles/access and alerting the farmer would be of extremely high value. The primary constraint being the rural nature of farm sites, which tend to have lower internet connectivity rates, and slow connection speeds due to the sparcity of the population.  Abilities to compress the data, or process locally and sending only alerts and photo captures of “unauthorized access” would ideally suit the unique constraints of remote sites.

Imagery and it’s impact to the bottom line is still an emerging connection in farming tech, but there are some start ups making exciting progress in this space, and we look forward to seeing more development to meet agriculture needs.