Our core capability. We draw precise polygon boundaries around individual plants, weeds, and field features — giving your model pixel-accurate ground truth data. Polygon annotation is the gold standard for training plant detection and segmentation models where shape and boundary matter as much as location.

Built for agricultural robotics teams, agri-tech startups, and computer vision researchers who need high-quality annotated field imagery at scale. Our annotators are trained on real Dutch farm imagery and understand the visual nuances of plant disease, growth stages, overlapping canopies, and variable field lighting — the hard cases generic annotation services get wrong. Every batch is reviewed by a Senior Annotator before delivery, with an average QA score of 0.97.

Fast, scalable object detection labels. We draw precise rectangular bounding boxes around target plants and weeds in agricultural field images — ideal for detection tasks where location, presence, and count matter more than exact boundary shape.

Our primary active work is weed detection for autonomous laser weeders — annotating chicory, carrot, and onion fields where the machine must distinguish crop plants from weed intruders at millimetre precision. We annotate to your exact specification — label taxonomy, overlap rules, and size thresholds are all configurable.

Precise single-point labels placed at anatomically meaningful locations — the base of a weed stem, the lower meristem, the point where a plant emerges from soil. Keypoints tell a robot not just that a plant exists, but exactly where to act on it.

Our keypoint work is done alongside bounding box annotation for autonomous laser weeding projects. Each weed receives a tight bounding box and a keypoint at its base — the coordinates the laser uses to target the plant with millimetre precision. Accuracy here is not cosmetic: a misplaced point kills a crop plant.