Agritech
Crops, weeds, yield and livestock, labelled for field ready models.
Annotation for crop health, yield estimation, precision farming and livestock monitoring across field and aerial imagery.
From seedling to harvest, and the herd
Every part of the workflow where we can take work off your team.
Crop health
Precision farming
Yield
Livestock and forestry
We work with field images, drone, multispectral and video.
What we label
Every class is written into a guide with your team before the first batch. Files come back as COCO JSON, YOLO, Pascal VOC, CSV, JSONL or your own schema.
Field and robot images
Plants
Bounding box and polygon
Crop, weed species, stem point
Disease
Polygon and severity
Blight, rust, mildew, leaf spot
Fruit
Bounding box and keypoints
Count, size, ripeness stage
Drone and multispectral
Fields
Segmentation
Row, gap, waterlogging, lodging
Forestry
Polygon
Canopy, clearing, burn
Video
Livestock
Track ID and keypoints
Animal ID, posture, gait, feeding
How we test the model
The tests that matter for agritech models, and what you get back from each.
Accuracy by growth stage
Model output reviewed across growth stages and seasons.
You get backAccuracy by stage and crop
Count error
Fruit and plant counts checked against reviewed images.
You get backCount error per image and block
Species confusion
Look alike species reviewed as their own set.
You get backConfusion between species
Where agritech teams start
Problems teams bring us, and the first piece of work we would scope.
Weeds and seedlings look the same in the first weeks.
What we doSpecies labels by growth stage, with a guide your agronomists review.AnnotateA yield estimate is only as good as the fruit count behind it.
What we doCounts and sizes labelled through leaf cover, checked block by block.AnnotateYour model learned one season in one region.
What we doNew seasons and regions labelled as they arrive and reported separately.Evaluate & alignRelated work
Twelve times the volume in five weeks, with quality holding at 95 to 98%
Human in the loop support for visual deviation detection on production lines, with a response time requirement under two minutes.
24/7 live review for production line visual AI, quality above 95% by month two
2,000 robot videos described every week by a team built in seven days
Half a million images for autonomous navigation, with every file kept on EU servers
Client and volumes [REAL DATA: Client names and volumes]
Bring us the agritech data your model gets wrong.
Send a small batch and get it back done, free, usually within 24 to 48 hours. You judge the work before you pay for anything.