CASE

Counting and sorting complex objects

Project: Automated sorting of broiler hen carcasses in order to identify possible rejects.

Implementation: Processing 6 carcasses per second, the analysis of camera input (taken at 50 frames per second) first determines each individual chicken; then, the input is analyzed further focusing on the legs and the general carcass, and a decision is made about the wholeness of each chicken and whether it has any defects or not. The entire evaluation history is logged and stored in the database for as long as the client wishes. This system helped increase the conveyor performance up to 180,000 carcasses per shift.

Value: Quality control is automated and devoid of potential corruption.

CASE

Counting and sorting complex objects

Project: Counting live chickens on a conveyor belt

Implementation: A high-speed camera (50 frames per second) is set up above the conveyor belt that is used to transport live chickens.With the help of a specially trained neural network, search is carried out for chickens within each frame; then an additional mathematical calculation is performed to determine the number of chickens in a huddle in cases when it is hard to single them out; any movement of chickens from one frame to another is also detected to eliminate the possibility of counting the same bird more than once.

Value: Complicated mechanical counting has been simplified multifold, which allowed increasing production and reducing the share of faulty products and the degree of inaccuracy.