The challenge

In an assembly line moving at high speeds, it is possible that due to quantity and speed, defective olives would be missed when being monitored by humans. 

By utilizing an object detection model, it is possible to detect visual defects in the olives more consistently, without limiting the speed or quantity. This will improve the quality of products being provided to the customers with minimal changes to the already existing production system.

Objectives

The goal is to detect defective olives while they are being processed. This will allow for a better way to accurately price a batch of olives. If the defective olives are removed, better quality of products can be provided to the customers.

How have we done it?

• Computer vision
• Object tracking and recognition
• YOLOv7
• Label Studio

The results

The model deployed was able to accurately detect when a defect was present. This allowed for better estimations of the average value of the yield of the products being measured.