Automated seed sorting with indexed carousel and machine vision
A cutting-edge seed sorting system using an indexed carousel, machine vision with CNN, and automated segregation. Optimized for maize and adaptable for various crops, this system enhances sorting accuracy and speed, aiming for industrial-scale application.
Overview
The proposed solution is an advanced seed sorting system that integrates an indexed carousel with machine vision and automated decision-making. Designed to enhance efficiency and accuracy in seed classification and segregation, this technology addresses both the singulation and classification challenges faced in traditional seed sorting methods. The system is particularly optimized for maize but is adaptable for various crops, offering a versatile solution for seed processing industries.
Technical specifications
- Singulation Carousel: Features an angled mirror imaging bay for simultaneous dual-side imaging, eliminating the need for flipping and ensuring continuous operation.
- Convolutional Neural Network (CNN): Trained to classify maize kernels with 96% accuracy for the R1-nj marker, processing at a rate of 100 seeds per second. The system is adaptable, allowing for different crop kernels by swapping the CNN model.
- Automated Segregation: Seeds are automatically ejected based on CNN classification, facilitating targeted sorting into appropriate streams.
Technology readiness level
Currently at Technology Readiness Level 3, the system is in the proof-of-concept prototype stage. Vineland Research and Innovation Centre is working on integrating and testing the system components to evaluate overall performance, throughput, and sorting accuracy.
About Vineland Research and Innovation Centre
Vineland Research and Innovation Centre is an independent, not-for-profit horticultural research organization in Vineland Station, Ontario, with a team of 100+ people focused on translating science into market-ready solutions. Its integrated campus combines pre-commercial greenhouses, advanced pathology and soil/substrate labs, a prototyping hangar for robotics and sensors, dedicated sensory and food technology suites, and field-scale research farms. A research-park setting with co-located government, academic and industry tenants enables on-site collaboration, testing and demonstrations. Vineland is funded in part through the Sustainable Canadian Agricultural Partnership, with support from Agriculture and Agri-Food Canada and the Government of Ontario. Commercialization is supported via defined licensing pathways and an active intellectual property portfolio.
delivered to your inbox.