By: Or Rotem, PhD Product Manager Seed-X Technologies
Seed-X Technologies R&D team had its “Robust seed germination prediction using deep learning and RGB image data” article published in the open-access, peer-reviewed “Scientific Reports”, one of Nature Portfolio most cited journals. It describes in great detail how GeNeeTM Technology, the brains behind the first ever AI-powered seed sorting machine ‘GeNee Sorter’, facilitates seed classification by specific quality attributes like never before, by mere image analysis.
GeNee Technology crunches plain image data to extract the quality grade of every seed by bespoke quality attributes. This is achieved by training GeNee proprietary algorithms per use-case to correlate between the image data and the quality attribute of interest.
This paper focuses on seed germination fate and addresses two main issues: Prediction universality and adaptiveness. Concerning the former, it shows how both seed germinability and germination vigor are predicted to render dozens of disqualified seed lots of multiple crops and varieties from different seed companies across three continents industrially appropriate, by bringing them to ≥90% seed usability rate with minimal unintended seed loss. With this, the paper demonstrates Ge-Nee Technology competence to handle multiple genetics, breeding programs, cultivation conditions and production pipelines.
The study further shows that training data adaptation enables to generate superior predictors and thus improve prediction performance. As proof of concept, it demonstrates how extensive crop-level data can streamline the sorting process by circumventing the need to train the system per new seed lot.
Collectively, this study illustrates how a single sorting event by the GeNee Sorter outcompetes the separation procedures conducted today in terms of germination rate improvement and seed recovery. Additionally, GeNee Technology grants the users with the ability to pre-define the quality and quantity of the sorting output, in accordance with their particular success criteria.
GeNee Technology has been used by seed companies globally since mid-2020 to improve the germination and genetic purity rates of commercial lots. So far, over 300 seed lots of various varieties and crops, including (and not limited to) Solanaceae, cucurbits, brassicas and ornamentals, have been processed and marketed (see picture). Processing by Seed-X Technology not only maximizes profitability for seed companies, but also allows them to minimize inventory costs and optimize production planning. fate or genetic purity, and it provides valuable insights for various applications. As mentioned, the technology is not limited to germination fate or genetic purity, and it provides valuable insights for various applications.
Seed-X team keeps applying the technology on various crops and use cases, upgrading seed lots, and improving classification performance by perfecting algorithm code and adapting the training data for every use-case. The pie chart presents the amount of seed batches upgraded since 2021 by GeNee Technology per crop.
Seed-X Technologies was founded in Israel by a group of seasoned hi-tech business experts and scientists in 2018. It has since expanded its activities across three continents: US, Israel and Europe (Netherlands). The company will be presenting its GeNee Sorter at the ISF World Seed Congress 2022 in Barcelona (booth 38-39).
You are welcome to drop by and see it live.
Scientific Reports (Volume 11, Article Number 22030)
