Generation
Labelled scenes at scale
Omniverse Replicator scenes with exact masks and boxes.
Learn · Synora
The hardest cases are the rarest: a new pest, a disease strain, a new crop, low sun through dust. Synora generates photorealistic, labelled scenes of them so Folira and Targis are ready before the field shows them.
ExampleBatch sizes are illustrative.
Capabilities
Generation
Omniverse Replicator scenes with exact masks and boxes.
Randomisation
Sun angle, soil, residue, occlusion and motion blur are varied.
Cold start
Starter data packs before a new region's first season.
Validation
Synthetic-trained models are scored on real held-out images and Acreon outcomes.
World models
Cosmos-based variation of real scenes (planned).
Licensing
Datasets and pipelines for equipment makers (planned).
How it works
1
Acreon and Folira show which cases fail.
2
Plants, weeds and pests are placed in randomised fields.
3
TAO fine-tunes the model on mixed real and synthetic data.
4
The new model must beat the old one on real images.
Specification
| Item | Planned |
|---|---|
| Scene engine | Omniverse Replicator |
| Labels | Instance masks, boxes, classes |
| Compute | DGX or cloud GPUs |
| Asset library | Crops, weeds, pests, soils |
| Validation | Real held-out sets |
| Output | COCO-style datasets |
Status
Nothing on this page is shipping today. Items marked planned are on the draft roadmap.
Place in the loop
Synora is the learn step. It reads what Acreon and Folira say is missing, makes data for it and hands better models back to Folira and Targis.
Questions
An agronomist or an engineer reads every message, not a sales queue.
No. It fills gaps. Every model is still tested on real images before it ships.
Licensing to equipment makers and ML teams is planned after the first packs ship.
By measuring: synthetic data is kept only if it improves results on real held-out images.
Next step
Send us one pass. We run it through Cropite and send back a per-plant report, labelled as an early example.