John Deere Acquires Orchard AI to Connect Crop Scans With Farm Equipment
Deere bought the San Francisco startup whose vehicle-mounted AI cameras count every fruit on every tree. The data will feed directly into tractor and sprayer displays.

By Clara Tan, Artificial Intelligence
3 min read
What changed
John Deere announced on 19 November 2026 that it has acquired Orchard Robotics, the San Francisco startup that operates as Orchard AI. The deal brings Orchard's vehicle-mounted camera systems and FruitScope software platform into Deere's precision-agriculture stack. Terms were not disclosed.
Orchard's cameras mount on existing farm equipment, tractors, sprayers, ATVs, and capture hundreds of images per tree as they pass through orchards and vineyards. Computer-vision models count every fruit, measure size and colour distribution, estimate yield per tree, and map canopy health. Until now, growers viewed that data in Orchard's FruitScope web and mobile apps. Deere says the plan is to pipe those per-plant insights straight into its Operations Centre, in-cab Gen 4/Gen 5 displays, and the John Deere Operations Centre Mobile app so that sprayers, harvesters, and variable-rate controllers can act on the data in the same pass.
"Orchard Robotics adds AI cameras and crop-level analytics directly onto Deere tractors and sprayers, feeding real-time yield and fruit data into in-cab displays and mobile tools," Deere said in its announcement. "That ties neatly into Deere's existing Operations Centre and See & Spray technology."
What Orchard AI can do
Orchard's stack has three main pieces: FruitScope Vision (the camera hardware and capture pipeline), FruitScope OS (the cloud processing and analytics engine), and Canary AI (the decision-support layer that turns per-tree measurements into variable-rate prescriptions).
Example 1: Apple yield mapping in Washington State
A 1,200-acre apple operation in Wenatchee, Washington, ran Orchard cameras on two sprayers during the 2025 growing season. The system captured 4.3 million tree-level images across 14 varieties. FruitScope OS returned per-tree fruit counts with a mean absolute error of 4.2% against hand-counted validation plots (n=240 trees). The grower used the yield maps to adjust bin allocation and labou [stored copy trimmed 6013 chars]
This story was first published by TechWhoKnows on 22 November 2026. Original: https://techwhoknows.com/tech/john-deere-acquires-orchard-ai-to-connect-crop-scans-with-farm-equipment
Sources
- 1John Deere press releasedeere.com
- 2Yahoo Finance coveragefinance.yahoo.com
- 3Global Agriculture reportglobal-agriculture.com
- 4Orchard AI websiteorchard.ai