World class algorithms

Fizyr applies deep learning to enable automated picking of unknown items for years. Our algorithms provide over 100 grasp poses each second, including classification to handle objects differently, but also perform quality controls and detect defects to prevent damaged items from being placed on a sorter.

Fizyr algorithms provide all relevant information, including:

• Segmentation (incl. white on white flats or reflective bags)
• Classifiers of type of parcel (incl. box, bag, envelope/flat, tube, cylinder, deformable, etc.)
• Determining outliers or non-conveyables
  (i.e. damaged goods)
• Best possible grasp poses in 6 DoF
• Multiple ordered poses per object

Robust object detection in various conditions 

Fizyr’s computer vision software successfully detects a large variety of objects for multiple logistics applications. Using a proper stereo camera for triangulation, our algorithms properly segment, classify, and propose the best possible grasp poses. 

Trained with a unique dataset with millions of images of challenging logistics environments, our algorithms can also provide valuable information on damages, package material and slip sheets which are relevant at picking cells in harsh logistics environments.