The Flex Sorter family is a range of advanced sorting machines. These machines can be custom-configured to meet specific demands on functionality and capacity.
Major features:
- Multiple imaging stations
- High-resolution RGB (default)
- X-Ray (optional)
- Chlorophyll fluorescence (optional)
- Hyperspectral NIR (950nm .. 1650nm)
- Wide range of crops (from celery to squash)
- Capacity from 10 seeds/sec ... 75 seeds/sec (more on request)
- Very accurate blow-off mechanism. < 1% at 50%/50% separation
- Sorting recipes based on
- AI engine (see below)
- Classical parameters (color, shape, etc)
- Smart adaptive feeder, for controlled seed separation
- Optional real-time blow-off error monitoring
- Optional 1-1 sowing unit. For sowing seeds for growing trials (see below)
AI-engine
The AI engine is used to discriminate a seed into the "accept" or the "reject" category.
The AI engine requires a sorting recipe, called the "trained model". Models can be loaded in ONNX format. These models can be created as follows:
- Using the DC-DL software application of SeQso
- Using your own or 3rd party software for training Deep learning models
If the sorting machine has multiple imaging stations, the AI engine can use all images of a seed to make the decision.
 |
|
