Deep learning machine vision software for chicken wings
Chicken wings control system

Industrial vision in deep learning for the food industry

Deep capture allows the identification and detection of several defects classes in real time at a rate of approximately 5000 parts / hour. The deep learning machine vision system is integrated on a conveyor at the output of an overhead cutting and winding system.

Example of the faults classes detected:

  • Hematomas and haemorrhages
  • Presence of the whip
  • Feathers
  • Broken bones
  • Badly rolled up wings

The subject complexity lies in the products shapes and appearance variability as well as certain defects are subject to interpretation.

Consumer PC

INTEL CORE I9 7940X
32GB RAM
2x GEFORCE RTX 2080TI-11G
WINDOWS 10

SVS VISTEK camera

Configuration of the deep learning model compatible with Google’s open source platform: TensorFlow.

Results detection and communication on Deep Capture software.

Taking photos and labeling on deep vision software