Semios: Helping growers produce more sustainable, profitable crops

About Semios

Semios develops precision crop management solutions for the agricultural industry. The company provides a proprietary system of in-crop wireless networks coupled with remote sensors, real-time pest monitoring, and variable rate biological pest control.

Industries: Agriculture, Forestry & Fishing, Technology
Location: Canada

Tell us your challenge. We're here to help.

Contact us

About Pythian

Ottawa-based Pythian is a Google Cloud partner with specializations in the areas of Machine Learning, Cloud Migration, Data Analytics, and Infrastructure.

With Google Cloud services such as BigQuery and Google Earth Engine, Semios helps growers produce more sustainable crops through precise, granular, real-time data.

Google Cloud results

  • Gathers data from 500,000 IoT sensors across 80,000 acres
  • Integrates data from first- and third-party sources to gain insights into crop risks
  • Identifies pests via machine learning, helping growers to reduce the population of a moth by about 1.5 billion in 2018
  • Integrates geospatial with sensor data for a holistic view of orchard risks

Processing 166.4M data points daily

Semios is a precision farming platform for growers of high-value crops, including almonds, pistachios, and apples. The company's mission is to help growers produce greater yields while making the production of these crops more sustainable by reducing farmers' needs for water and pesticides. It's an important goal, given that agriculture uses up to 70% of the world's fresh water, according to The World Bank. And the industry is pressured by the possibility that the drinking water for 50 million people in the United States is potentially contaminated by pesticides, according to estimates by the U.S. Department of Agriculture.

To achieve its mission, Semios relies on a network of 500,000 installed Internet of Things (IoT) sensors across 80,000 acres of orchards. The sensors gather big data related to insects, moisture, weather, localized network performance, and other elements and transmits the data via a local gateway to an LTE cellular network and from there to Semios servers. The end result is a service that gives growers a simple, intuitive dashboard from which they can easily assess and respond in real time to insect, disease, and plant health conditions while finding opportunities to reduce water and pesticide use and improve crop quality.

"Agricultural insights tend to exist in siloed databases. BigQuery allows us to connect various data sources where no connections previously existed. We were the first company to bring it all together, to give our customers the big picture of their crops."

Tessa Trethewey, Director of Strategic Partnerships, Semios

Agriculture is the largest industry in the world, according to the World Wildlife Fund. Yet, until recently, optimizing crops has mostly involved growers' tapping into their experience, intuition, and time-honored best practices. For example, growers may spray crops with pesticides every two weeks as a preventative measure and hope for the best, rather than spraying only when and where pesticides are truly needed based on specific, real-time data that pinpoints an exact insect infestation problem. "Using data enables growers to wait for a risk to materialize and then take action on it, which ultimately makes agriculture much more sustainable and less costly," says Dr. Michael Gilbert, CEO, and Founder of Semios.

In its early days, Semios used a PC to collect sensor data and host it for customer access. As the number of networked sensors grew and scalability became an issue, Semios moved data ingest, storage, and processing to the cloud.

"We needed a robust cloud solution that could scale easily as we grew, would give us much greater reliability in continuously capturing sensor data, and would translate all that raw data into information growers could receive in real time on their smartphones, so they could understand right away what they need to do to reduce their risks," says Michael. "Google Cloud checks all those boxes and was the best fit for our needs."

Giving growers the big picture

From its remote sensors and devices, Semios collects approximately 200 million data points daily, which is expected to more than double in 2019 as the company grows and its datasets, including imagery, become more complex. The data Semios collects currently encompasses weather, including temperature, humidity, rainfall, and wind; images from networked camera traps, which capture and identify insects and help determine population pressure; network stats such as signal strength, number of hops, and more; and pheromone dispenser performance, including puff count and spray sensor readings.

"Knowing the best time to spray can mean using less pesticide or even a non-toxic alternative in a more targeted area. This is much better for growers and ultimately for consumers. This can lead to lower costs for growers and respond to growing consumer pressure to reduce the use of chemicals in food production."

Dr. Michael Gilbert, CEO, and Founder, Semios

The company uses BigQuery to understand how its networks of sensors are performing, so it can make adjustments as needed. Semios also collects third-party data related to pesticide and fertilizer application, crop yields, and pest imagery and brings these various datasets together in BigQuery. "Agricultural insights tend to exist in siloed databases," says Tessa Trethewey, Director of Strategic Partnerships for Semios. "BigQuery allows us to connect disparate data sources where no connections previously existed. We were the first company to bring it all together, to give our customers the big picture of their crops."

Micro climate data related to the crops is pushed to Google Cloud machine learning algorithms that help growers predict risks to their crops. For example, pests are a huge threat. Consultants are traditionally hired by growers to count catches in insect traps, scout for disease, and report this information back to the grower on a weekly or bi-weekly basis. Their challenge is conditions can change on a daily basis and reporting on a weekly or bi-weekly basis isn’t ideal for controlling pests.

Semios installed small digital cameras — the kind found in smartphones — inside insect traps in customer orchards. The cameras take multiple pictures every day, but it was impractical to check those images manually. To address the challenge, Semios adopted the Cloud Machine Learning Engine to train specialized AI models to identify the type and number of pests photographed inside traps.

With the Cloud Machine Learning Engine, Semios helps growers identify exact pest threats in real time. "Knowing the best time to spray can mean using less pesticide or even a non-toxic alternative in a more targeted fashion," explains Michael. "This can lead to lower costs for growers and respond to growing consumer pressure to reduce the use of chemicals in food production. Almond growers can suffer incredible economic loss every season caused by a pest that not only damages the nut, but can also introduce a fungus that produces a dangerous toxin. A data-driven approach helped our growers reduce the population of a moth by about 1.5 billion in 2018 meaning fewer damaged nuts and improved safety and quality of food available to consumers."

"With Google Earth Engine, we can bridge granular data from ground sensors with data from aerial sources in a way that hasn't been done before. We can give growers the most accurate insights into plant yields, health, and irrigation, so they can make the best decisions to optimize their crops."

Tessa Trethewey, Director of Strategic Partnerships, Semios

Semios adopted the open source machine learning framework TensorFlow, created by Google, to build its AI capabilities for insect monitoring. The company also leverages Cloud Functions for making real-time prediction requests to the model deployed to the Cloud Machine Learning Engine.

Accomplishing the previously impossible with Google Earth Engine

Most recently, Semios engineers have become early adopters of Google Earth Engine, a service enabled by Google Cloud that combines satellite imagery, geospatial datasets, and planetary-scale analysis capabilities. Semios uses Google Earth Engine to connect geospatial data with the data gathered by Semios sensors on the ground.

"A lot of agricultural decisions are now being driven by imagery from satellites and drones," says Tessa. "With Google Earth Engine, we can bridge granular data from ground sensors with data from aerial sources in a way that hasn't been previously done. We can give growers the most accurate insights into plant yields, health, and irrigation, so they can make the best decisions to optimize their crops. Because Google Earth Engine is part of Google Cloud, it fits easily into our normal data flow, which simplifies how we can draw conclusions from the data to share with growers."

Scaling without focusing on resources

Auto scaling capabilities in Google Cloud services such as BigQuery are enabling Semios to grow rapidly. "As an entrepreneur I want to focus on our intellectual property, and on hiring more soil scientists and entomologists," says Michael. "I don't want to focus on infrastructure, and I don't want to hire people whose job is to worry about that. We know Google Cloud will take care of it for us, and that frees us to create value for customers much faster."

Working with Pythian and Google

Google partner Pythian was helpful in transitioning Semios to Google Cloud, enabling the agricultural startup to devote its time and resources to building its software offerings.

Google team members worked closely with Semios to help the company incorporate Google Earth Engine into its data workflow. "With Google Earth Engine, BigQuery, and machine learning, the innovation at Google is one of the biggest reasons we wanted to partner with them," says Tessa. "The strong support they've given us has really cemented the relationship."

Tell us your challenge. We're here to help.

Contact us

About Semios

Semios develops precision crop management solutions for the agricultural industry. The company provides a proprietary system of in-crop wireless networks coupled with remote sensors, real-time pest monitoring, and variable rate biological pest control.

Industries: Agriculture, Forestry & Fishing, Technology
Location: Canada

About Pythian

Ottawa-based Pythian is a Google Cloud partner with specializations in the areas of Machine Learning, Cloud Migration, Data Analytics, and Infrastructure.