Goal of the system was to use deep learning object detection to sort watermelon seeds (falling under gravity) based on their ploidy (2x, 3x, and 4x). This case, the 3x seeds were considered the rejects.
- Systems grabs data of falling seeds using 2 highspeed sentech cameras connected through a Euresys frame grabber
- Using a deep learning objection model (YOLOv5n), detect seeds based on ploidy
- Eject the 3x seeds using solenoid controlled ejectors (pressured air)