Although we’ve been planting crops for centuries, farmers are plagued with the same critical and potentially game-clinching decision each season: Which seeds should I plant?
They navigate this question based on their region, weather conditions, yield, and other key variables. Over the decades, seed companies have largely focused on enhancing yield, but with climate change making weather more severe and increasingly unpredictable, farmers are now needing seeds that cover the gambit.
“Conventional seed selection is very wasteful and hit and miss,” Sarel Ashkenazy, founder and CEO of Seed-X, told AFN. “Our technology supports the agricultural ecosystem by enabling breeders and producers to achieve better results that are faster and more sustainable. They can achieve the traits they desire at a fraction of the cost, while significantly reducing the use of agricultural inputs like land, water, fertilizers, energy, and chemicals.”
Seed-X, which won Most Innovative International Startup Pre-Series A in the 2019 AgFunder AgriFood Tech Innovation Awards, uses advanced machine vision technology and customized deep learning algorithms to revolutionize seed breeding and production cycles. The startup identifies genetic traits of seeds and crops based on phenotype features and is hoping to use the big data it gleans in the process to transform the entire industry.
By Ashkenazy’s account, Seed-X is the only company using computer vision and AI at the seed level to detect genetic characteristics all the way down to the specific genetic traits.
We caught up with Ashkenazy to learn more about how Seed-X is trying to revolutionize the seed selection process.
Can you explain how your technology works?
Our technology is based on a combination of computer vision, artificial intelligence and proprietary algorithms that analyze the seed’s phenotype to detect genetic characteristics or traits on the seed level. The breakthrough is that this approach is non-destructive to seeds and allows genotype analysis without doing molecular genetics tests.
The first product we launched is called the GeNee Breeder, a portable seed analysis tool that helps plant breeders perform the selection process for every breeding cycle with greater efficiency. The main benefits are shortening the breeding time and increasing the probability of success. Our product pipeline features sorting machines for seed production that will enable sorting by genetic purity, germination probability, and health parameters.
Who is your customer? Has it been difficult marketing to them?
Our primary target customers are seed companies. Most people working in plant genetics do not really believe that we can “see” plant genomics on the seed level. So, naturally our challenge is to tackle this skepticism barrier and we are doing this by conducting proof-of-concept experiments to demonstrate that it really works.
What would you list as the three biggest contributors to your success so far?
Many years of proven experience in the field of computer vision and AI algorithms for phenotype analysis, the fact that we are originally not from the agriculture industry, and a great team of people including strategic advisors. Seed-X is part of a group of companies with long background in computer vision and AI. one of the founders founded face.com that was sold to facebook a decade ago. He also founded FDNA also related to facial recognition but together with the correlation to the human DNA, that is how Seed-X idea came, to correlate between the Seed “Face” to its genetics.
The fact that we are not coming from this field allowed us to be open-minded to check everything without thinking about possible limitations or impossibilities.
What are three challenges that you’ve faced along the way?
Overcoming the initial skepticism of some customers. For some of our abilities, such as classifying hybrid or detecting specific traits like virus resistance. Most of the breeders and geneticists could not believe that we can make the detection without using any molecular testing or PCR, which is a specific molecular genetic testing method. It is challenging to convince them that it is doable but once they see the successful proof-of-concept on their genetic material, the skepticism barriers break.
Developing the precision and capabilities of our algorithms. In order to train any AI algorithm and specifically using the deep learning that we use, you need to have a large data set for the training. Because the industry could not compromise on the results it was challenging in the past to get high results with a low training set. Currently, after 2 years of operation and making more than 150 experiments with different companies and academic institutions, we are not facing this challenge anymore with most of the crops.
Scaling up R&D. In order to overcome the skepticism, we had to perform lots of experiments in a relatively short time. In the beginning, it was challenging because we approached many companies simultaneously and had to do many experiments on different crops at the same time.
What is one thing you wish consumers understood better about food, farming, and technology?
AI agriculture is not something to be feared but embraced. The use of AI can help produce higher quality food in a more sustainable, affordable, and less wasteful way. AI technologies are creating new ways, which were not known until today, to analyze existing data. All areas of science related to plants biology, pathology, seeds, and grains physiology can make huge steps due to AI.
What advice do you have for other startups out there?
In the agtech industry, there’s no substitute for being patient and going step by step. There are no shortcuts to building credibility and trust among the seed companies. Therefore, make sure to raise enough money to be able to complete your planned set of experiments and develop your technology.
Where do you hope to see your company in the next three years?
Making a big change in processes and changing lots of old methods. If you can sort out bad seeds from a batch, not only can you enhance every batch, but the industry will not have to compromise on quality anymore. There will be no more 90% germination as acceptable or 98% purity. All batches sold will be 100%. Moreover, we believe that once we will have the capabilities to enhance every batch, the seed production methods will change and will be less labor-intense as of today.
Guest article: AI can transform precision agriculture, but what are the legal risks?