Business background
- In the Baltics, post–ice age geology has left large volumes of mixed, oval-shaped rocks
- Two common materials are dolomite and granite, often similar in size and appearance
- Dolomite can be up to 5× cheaper, so mixing materials directly reduces product value
- It’s commercially important to extract granite pieces with good shape and visual quality
- Manual sorting is slow, inconsistent, and depends on operator experience
What this demo demonstrates
- A solution developed from the ground up around the customer’s own definition of valuable rocks and acceptable shapes
- A custom-trained network created specifically for this task (not a generic off-the-shelf model)
- Reliable classification under real conveyor-line conditions
- Seamless integration into production lines using standard industrial communication protocols
Benefits for end users
- Higher margins by separating low-value material early and keeping granite streams clean
- More consistent output quality and fewer customer complaints/returns
- Reduced labor load and less dependence on individual operator skill
- Fast and predictable automated decisions suitable for real-time sorting equipment
What AgirVision can deliver for your use case
- Translate your expert criteria into a clear labeling and training workflow
- Train a task-specific model optimized for accuracy and speed on your target hardware
- Industrial integration via Modbus, CAN bus, OPC-UA, Ethernet/IP, or custom machine interfaces
- Full system deployment from camera input to actuator control or SCADA reporting
Want something similar?