High Speed Determination of Tulip Bulb Pose Using Computer Vision

The cost of labour makes up almost a third of the operating costs for greenhouse, nursery and floriculture operations. Many growers are seeking ways of automating labour intensive processes to reduce their exposure to increasing minimum wages and shortages of available labour. Planting flower bulbs is an example of one such process. The main barriers to developing an automated system are: identifying the bulbs with computer vision despite significant variations in size and shape within flower species; determining the bulb pose so that the bulbs can be planted pointed end up for consistent growth; operating fast enough to replace multiple workers and stay cost-effective; and visually checking for bulbs that are rotten or moldy. The researchers will need to evaluate existing sensors and computer vision techniques, and then adapt them to perform this task. The partner organization will benefit from the project through the potential commercialization of the developed system.

Yahu Choudhary
Faculty Supervisor: 
Gary Bone
Partner University: