According to the weather forecast, it is expected to be warm tomorrow. Je Hoon, a new farmer who grows perilla leaves, thinks it would be fine to increase ventilation of his greenhouse by opening the windows for the rest of the night. He is not sure, as he lacks farming experience, if his decision to increase ventilation is a right one. He relies on educated guess while he knows that a wrong decision may lead to a negative effect on the perilla leaves, but since it is too late to ask his neighbors and fellow farmers, he decides to move on. After all, it’s been a long and hectic day for Je Hoon, and he wants to go home and sleep as soon as possible.


The next morning, Je Hoon finds out that a lot of his perilla leaves have withered. He feels guilty and regretful over opening the windows last night to increase ventilation. What he thought would be perfectly fine turns out not to be ok. He tries to think how ventilation affected various factors that led to the withering of his perilla leaves – low temperature, high humidity, high carbon dioxide concentration in the atmosphere, high radiation, etc.


Perplexed as to why his perilla leaves have withered, he searches the Internet. He also asks his neighbors, fellow farmers and agricultural experts. However, people have different opinions and thoughts because they don’t have reliable data to analyze.


Then, the Agricultural Technology Center introduces a system called, “G-Smart.” This is a greenhouse control system that determines the level of ventilation required based on data collected from Internet of Things devices implemented in the greenhouse such as temperature, humidity, gas level, etc. The solution recommends farmers to open or close windows to ensure optimal ventilation rate. Je Hoon thinks this is a perfect solution for him.


Impressed with what G-Smart has to offer, Je Hoon immediately calls the service provider and asks them to install it. The system is composed of different components – sensors, access to the Internet, auto ventilation system, IoT, cameras, etc.
The devices collect data from the greenhouse (temperature, humidity, gas level, ventilator status, radiation intensity, time of day). Je Hoon is amazed to see real-time data from his greenhouse on the mobile application connected to G-Smart.


Je Hoon can not only access data, but also authorize G-Smart to automatically fully open, fully close or open by half the greenhouse windows based on the system’s recommendation.


After Je Hoon has opened the greenhouse windows by half based on G-Smart’s recommendation, the mobile app now shows the new temperature, humidity and CO2 level inside the greenhouse as well as real-time surveillance of his perilla leaves. Je Hoon can also keep track of past data and pictures of his perilla leaves under “Details.”


A year after Je Hoon has started using G-Smart, he has had more leisure time as farming perilla leaves has become less time consuming. Consumers who purchase perilla leaves in supermarkets are also more satisfied with perilla leaves, as there are now fewer withered ones. Furthermore, the Agricultural Technology Center, which is responsible for promoting G-Smart, has played a pivotal role in revitalizing the local economy by enhancing perilla leaf production.

The Busan CIC is leading a team consisting of Geumsan County, Digilog (a Busan CIC startup members), and Dong-A University (DAU) to develop an automated, Artificial Intelligence-based greenhouse ventilation technology called “G-Smart,” which will help 2,184 perilla leaf farmers in Geumsan County improve ventilation inside their greenhouse, thereby enhancing their production efficiency.

Each smart greenhouse is equipped with a ventilation control system, which is composed of air vents and extractor fans. Operating this system affects not only the ventilation rate, but also the temperature, humidity, carbon dioxide concentration and many other factors that determine the growth of perilla leaves inside a greenhouse. The cultivation of high-quality perilla leaves demands a decrease in humidity following precipitation as well as a level of CO2 appropriate for photosynthesis, for instance. Unfortunately, controlling ventilation has been a difficult task for most farmers in Geumsan County, as they often do not know when and to what extent they should increase or decrease ventilation. In response, farmers try to search the Internet or ask other farmers, but the unreliability of these information only increase their anxiety. Additionally, they also find manually opening or closing air vents, and turning on or off extractor fans inconvenient and time consuming.

The Busan CIC created a team that comprised Geumsan County, Digilog, Dong-A University and others to launch the Smart Farm Challenge. Using the Amazon Working Backwards process, the Busan CIC delivered an Innovation Workshop as well as a Solution Workshop, and conducted end-user interviews to define the problem and develop a creative solution.

The prototype that the team is first developing is G-Smart’s mobile application that perilla leaf farmers in Geumsan County will use to automate ventilation. The mobile app is connected to sensors and IoT devices installed in the greenhouse, which will collect real-time environmental data that influence the growth of perilla leaves such as temperature, humidity and CO2 concentration. This data is then compared to images of perilla leaves, which can be analyzed to measure their growth rate and are taken by cameras moving on a rail inside the greenhouse. By examining the effect of a particular greenhouse condition (i.e. a specific level of CO2 concentration, humidity and temperature) on the growth of perilla leaves, the team will be able to quantify an optimal range of greenhouse conditions for perilla leaf growth. Finally, automatic ventilation will help to ensure that the conditions in the greenhouse fall within this range. G-Smart’s mobile app will recommend farmers to increase or decrease ventilation in the two following ways: fully open, fully close or open by 50% the air vents, or turn on or off the extractor fans. Farmers can then authorize the app to follow its recommendation, automating ventilation in the greenhouse. As more environmental data and perilla leaf images are collected, the optimal range will become narrower and thus more accurate, leading to a virtuous cycle drawn below.

As a result of this virtuous cycle, the team expects G-Smart to save 2 hours of farmers’ time every day and increase annual perilla leaf production by 5%.

Furthermore, as G-Smart accumulates more data, the team will be able to prioritize different environmental data to determine which are more crucial to the growth of perilla leaves. Data prioritization is important because farmers will need to use their limited budget as efficiently as possible. In the future, as G-Smart identifies less significant perilla leaf growth factors, the team expects farmers to spend 30% less on sensor and IoT device installation costs. By minimizing their spending on devices and enhancing their perilla leaf production, G-Smart will ultimately help farmers maximize their financial gains.

Before implementing G-Smart across Geumsan County, the team will test the prototype with 5 greenhouses and identify areas of improvement so that the team can make the necessary changes before distributing the solution to a larger farmer group. Once G-Smart is implemented across Geumsan County, the team hopes to introduce the prototype across greenhouses in South Korea.

The data that the mobile app uses as well as the IoT devices and sensors that collect these data are both currently from a third-party vendor. However, the Smart Farm Challenge team will seek to install its own devices and generate its own greenhouse data in the future.

As part of the solution design process, a fictional Press Release and Frequently Asked Questions document was used to create a vision of what the solution could accomplish.

"Safeguard of Greenhouse (G-Smart)"
Geumsan-gun distributing the first AI-based management system of greenhouses for sesame leaves
Dec. 2021 - Geumsan-gun (the Gun) announced that the Gun succeeded in developing "Safeguard of Greenhouse" (G-Smart) jointly with Amazon and the Digilog Team of Dong-A University in Busan, which is an AI-based Korean smart farm solution for sesame leaf farms. G-Smart is an AI solutions that resolves the ventilation issues, one of the hardships experienced by sesame leaf farms, and automatically monitors the health conditions of sesame leaves. The AI analyzes the best timing and frequency of ventilation of greenhouse and suggests proper management measures to efficient ventilation control. Further, the AI-based analysis of camera images allows us to easily monitor the healthy growth of sesame leaves. It is expected that G-Smart will provide assistance on monitoring of ventilation control and health conditions of sesame leaves as well as significantly reduce troubles of farms for frequent visits to greenhouse, and manually opening and closing windows of the houses.

The major control factors for cultivating sesame leaves mainly in greenhouses include temperature and humidity, concentration of carbon dioxide and opening/closing windows relevant to ventilation. The cultivation of high quality sesame leaves demands prevention of rising of humidity caused by rain and fog inside the greenhouse, and precise keeping of concentration of carbon dioxide essential for photosynthesis. If control of ventilation is performed poorly, various symptoms may take place such as fungal diseases and interruption of gas flow to significantly reduce the commercial values of sesame leaves, and in extreme cases, can lead to the withering of the whole leaf. G-Smart, based on AI technology, analyzes major environmental factors inside the greenhouse of sesame leaves, and automatically identifies the health conditions of the leaves. G-Smart further predicts environmental changes inside and outside greenhouse beforehand for optimal control of ventilation. The AI monitors the health conditions of sesame leaves, and analyzes and predicts the optimal conditions of environmental factors such as temperature, humidity and concentration of carbon dioxide to advise the opening/closing of the greenhouse and adjustment of exhaust fans allowing farms cultivate sesame leaves without difficulties on ventilation control.

Mun Jeong-U, the Governor of the Gun, who is also a farmer, said, "It is very good thing to provide the farmers of sesame leaves in the Gun with versatile control services through G-Smart. Farming based on scientific data analysis and AI, not on subjective feeling or assumption anymore, is such a situation that is significant. We just started the first step of G-Smart, and anticipate continuous development and progress of the system. In addition, continuous analysis of environmental data and growth data, and buildup of such data in AI cloud of the Gun will apply to other crops to promise the bright future of AI agriculture of the Gun."

Lee Sae-rom (56) in Chubu-myeon of the Gun said, "AI analyzing and providing suggestions regarding the complicated ventilation control of greenhouse frees me from some of the tedious daily chores. Also, cameras that automatically monitor the health conditions of the crops allow me to feel reassured. I had to visit the greenhouse for several hours every day to examine the crops for ventilation control, but now I can focus those hours on other tasks. I get two to three hours more a day which allows me to take my time while farming, but I am satisfied with the unexpected growth of throughput of sesame leaves by 3 to 5%. I like that I receive specifics of control continuously on my smartphone and will consider transitioning to automatic window opening and closing."

It is anticipated that G-Smart improves convenience of farms and precisely controls ventilation of greenhouses in the Gun to lead scientific and convenient farming of such farms. The Gun indicated that the Gun will be the leader of digital agriculture in Korea by distributing the smart farm of low costs and high capabilities to farms. Please visit the homepage of Geumsan-gun (https://www.geumsan.go.kr) or call the Agriculture Technology Center of the Gun (041-750-3584) for further information about G-Smart.

[FAQs from Customers]
Q.How much is the initial investment in G-Smart?
A. The services will be implemented by utilizing automation systems to the largest extent, if any at farms, and initial investment will vary dependent upon systems operated by farms. It is, however, expected that farms are not obliged to make payment of initial investment.

Q.Are there any additional systems for operating G-Smart? Is G-Smart compatible with sensors operated at farms? Which system will be used for environment analysis of the greenhouse?
A.Most systems operated by farms so far will be used as they are, but additional systems may be required such as temperature and humidity control systems and fans. Upgrading communication systems or part replacement may be required for facilitated cloud communication over Internet. Checking compatibility of systems and installation of additional communication systems will be performed by physical visits to farms.

Q.Are there any requirements for operation of G-Smart? Is it possible to directly install G-Smart on a greenhouse under operation? Are there any cases of infeasibility of distribution or installation of G-Smart?
A.In general, G-Smart can be installed and operated at places where Internet is connected, or allowed for a connection. There should be space inside the greenhouse for installing sensors and systems. When these requirements are satisfied, installation of G-Smart will take one or two days dependent upon situation of farms. You may install G-Smart with the help from an installation company introduced by the agriculture technology center.

Q.How is ventilation performed? Is information manually entered, or automatically recorded?
A.The environment inside the greenhouse and conditions of sesame leaves checked by cameras are automatically collected and analyzed.

Q.Will systems be free from failure at excessively low or high temperatures? Are after-sales services promptly provided? Do after-sales services incur any costs?
A.The systems, by default, are of IP6 Class thus they are water- and dust-proof, and stably operate in environments of high temperature and humidity. If any system fails during operation, it will be inspected within 24 hours. Free warranty period of hardware is two years after installation.

Q.I'm not familiar with the operation of a smartphone and computer. Can I still use the systems without a smartphone or other similar devices?
A.Basic operation of the systems is not allowed if you don't have a smartphone. You should be capable of using a smartphone; however, you may get help from the agriculture technology center, if you are not familiar with it. Services will be provided through voice guidance on smartphone in the future.

Q.Is it possible to install the systems again when the greenhouse collapses or gets torn by natural disasters (e.g., storm, earthquake and heavy snow)?
A.The re-installation of systems is possible; however, costs may incur to some extent.

Q.What is the countermeasure of the system for problems such as a blackout?
A.If Internet connection between farm and server is broken due to a blackout, such situation will be automatically notified via smartphone.

Q.May I receive the analysis results of ventilation on weekdays as well as weekends?
A.You can receive the results round the clock (may vary dependent upon software updating).

Q.They say there will be no restrictions of time and place. How much time will be saved on average?
A.Situations vary among farms, and it is hard to estimate saved time numerically; however, saving two or three hours a day is expected.

Q.How much do fixed costs incur for continuously receiving the services?
A.There is no initial cost for installation and purchase of systems, and the services will be provided for free for a specified period. The consistent incurrence of communication costs and service costs of AI-recommended automation is expected, and for further advancement, costs of ten to twenty thousand KRW may incur monthly. However, there is no plan of charging services at the moment.

Q.Who will be responsible for damage to crops caused by erroneous ventilation control after installation of the systems? Is compensation provided if all of sesame leaves die because of faults of system operation?
A.Such situation is expected to be rare. However, if damage to crops takes place due to system error, the Gun and the agriculture technology center will compensate such damage.

Q.Is reimbursement allowed if there is no significant change on convenience and throughput after using G-Smart?
A.Yes, reimbursement will be made.

Q.Are there methods for correction of collection of erroneous information?
A.Erroneous information will be manually corrected in the initial stage; however, the systems will be advanced for automatic correction of missed/erroneous data by machine learning and AI thereafter.

Q.Does the service include real-time monitoring function?
A.Monitoring at 5 to 10-minute lapse is possible.

Q.Is there a method for detecting abnormal operation of the system?
A.There may be various causes of malfunction of systems. Software trouble and sensor malfunction will automatically be detected and alarmed. Regular inspection will be provided every three months (e.g., regular inspection of water purifier) to monitor stable operation of systems round the clock.

Q.Are there any parts of services requiring of continuous management in standpoint of engineering?
A.It is expected that no parts are requiring continuous management at farms. What are the differences from services of other companies (e.g., Green Labs and Farms Cube)? Operating AWS cloud and edge-cutting AI technologies of Amazon is the difference from other companies.

[FAQs from in-house staffs/stakeholders]
Q.Are there methods of detecting problems at the agriculture technology center of the Gun?
A.Any troubles of hardware and communication error with greenhouse are automatically notified in letter to smartphone of staff of the agriculture technology center.

Q.Any troubles of hardware and communication error with greenhouse are automatically notified in letter to smartphone of staff of the agriculture technology center registered on the system. What is the dimension of system required for database buildup and AI technology?
A.There are no computers/server systems to be installed for the services at the agriculture technology center. Collection and processing of data and AI analysis are performed in the Amazon Cloud.

Q.How much is the launching price?
A.The Gun will provide support of budgets of 70 million KRW, and the farm will operate the system without any costs at the time of launching. Please, however, note that costs of ten to twenty thousand KRW may incur from communication and Amazon Cloud charge two years later.

Q.Will reorganization of system be required upon growth of the number of participating farms?
A.No significant reorganization is required even when number of participating farms grows. However, the initial investment of hardware may incur to some extent dependent upon circumstances of new farms desiring participation.

Q.What are the core benefits of the solution to the Gun?
A.The most significant benefit is the increase of income of the Gun farms resulted from improved productivity at farms. Another significant benefit is the expansion of free hours and rise of better life index of farmers of sesame leaf farms.
It is anticipated that the Gun will be represented as a major production site of high quality sesame leaves to consumers.
Collaboration with Amazon, a global IT company, will enhance the global position of the Gun. Cases of the Gun will widely be introduced worldwide through the website and AWS Featured Challenge of Amazon, and the Gun will become capable of demonstrating success cases and models of advanced agriculture.

Q.How much costs and time will be required for solution development?
A.It is expected that solution development will take two or three months, and the budgets of the Gun will cover the costs of development.
Q.What are the difficulties for launching the app?
A.Precision of analysis and prediction may not be high as expectation at the initial stage of services. Precision will gradually rise when accumulated data increases, and learning of AI model is improved accordingly.

Q.What are the initial target customer groups of the solution?
A.Five farms of fresh nutrient sesame leaves in the Gun.

Q.What is the success indicator of the solution?
A.The goal is an increase of sesame leaf throughput by 5%.

Q.How is the system operated and managed?
A.General data and know-how of farms of quality sesame leaves in the Gun will be obtained in the initial stage of operation with less data, and then services will be operated at other farms with the data. Once the data is built up to proper level, system will be operated with its own data. As for management, permanent staffs should be appointed, who complete education courses of operation of the system, server and data.

Q.What is the expansion plan once the solution is successful?
A.Continuous advancement of AI model will be performed, and the services will be expanded to price prediction and notifying optimal sales timing of sesame leaves. Also, the services will be expanded to farms desiring the operation of the solution.

Q.What is the legal requirement for collection of data from farms?
A.Consent on utilizing personal information (e.g., farm information) should be sought. The agriculture technology center of the Gun will provide the consent form.

Q.Are there any results of study on extent of window opening at certain situation?
A.It is planned to create test beds, and perform various experiments and studies at the agriculture technology center.
Test beds will be constructed by August, and sesame will be planted in September. Though the width and the length of greenhouses vary among farms, the space provides environment for cultivating and harvesting sesame leaves under diversified conditions. You can consider the greenhouses with total eight test beds with four different conditions of bed soil layers as follows: Pearlite 3 : cocopeat 7; Pearlite 4 : cocopeat 6; Pearlite 5 : cocopeat 5; and Pearlite 6 : cocopeat 4.

[Additional FAQs from the Gun]
Q.How is agreement executed for budgeting of the Gun?
A.The agriculture technology center is entitled to execute direct agreement for project budgets for supporting farms in amount up to 20 million KRW, and should invite bidding for budgets exceeding such amount. The participation and support of Dong-A University have not been facilitated as expected in the initial stage of the project, and amount of 20 million KRW or more should be paid to the industry-academic collaboration foundation of the university in the course of execution of an agreement with the foundation. Accordingly, the center drove execution of agreement with Digilog through bid invitation. Once bidding fails by adding restrictions on companies capable of joint development with AWS in the invitation, it is expected to execute agreement with Digilog without particular trouble. It seems that discussion on this issue is required; but it is also required to review feasibility of execution of service agreement with Digilog, not with the university. It is, however, worthwhile to consider that the project was commenced by Dong-A University. I will provide you with firm information after further review.