3 of 33

Introduction

The Problem

The Idea

Implementation

Impact

Precedent

Who we are

Lawn water waste crisis

Computer vision based sprinkler systems

Business Plan & Model

What effect it will have

How we know it will work

4 of 33

Introduction

5 of 33

Computer Science students who like to be outside

Clement Chatelain

Isabella DeBoer

Bode Packer

Delia Leonard

Jaxon Smith

6 of 33

The Problem

A huge amount of water is wasted on watering lawns

7 of 33

The Current Issue with Watering Lawns

Scarcity	Climate change makes weather patterns more extreme → drought in dry places
Cost	Water is a monthly expense Cost increases with scarcity
Societal Norms	Value of green space in cities Green lawns have long been seen as a sign of status
Knowledge	How can you best care for your lawn? Overwatering vs underwatering

8 of 33

Water Scarcity in Utah

Climate change contributes to drought:

Increased water temperature enhances evaporation which dries out soil
Warmer winters cause less precipitation during that time of year
Decrease snowpack and thus snowmelt

Overall, it adds stress to ecosystems, making dry areas (such as Utah) drier. ^[15]

Utah has felt this impact and will be the case study to demonstrate the impact of GreenSight.

^*9

9 of 33

“As much as 50 percent of water used for irrigation is wasted due to evaporation, wind, or runoff caused by inefficient irrigation methods and systems.” ^[2] Overwatering adds this unnecessary water to the environment, harming the plants and wasting water.

In 2022, Deseret News and The Hinckley Institute of Politics released a poll inquiring how often individuals water their lawn. The results to the right found a whopping 70% of people overwater despite major efforts to reduce that number.

Overwatering and underwatering often look very similar and people tend to assume the latter. This creates a positive feedback cycle that results in water waste.

A lawn only requires water once a week to stay healthy! ^[18]

10 of 33

Why don’t we just stop watering lawns?

The issue of water waste on our lawns has long been around, with no clear resolution.

“The use of lawns in our modern society is seen as a product of our lifestyle.Today, lawns cover a significant part of all green open spaces in cities (up to 70–75%).” ^[10] Citizens place high value on these areas to conveniently get outside for recreation. “The concept of the grass lawn as a status symbol originated with English aristocrats in the 17th century,” so it has been common practice for hundreds of years. ^[21]

Despite the increase of population in Salt Lake, the general consensus is that people don’t want to remove their grass just so that other people can move to the area. ^[16]

In sum, it has simply been proven to be unsuccessful.

11 of 33

The Idea

Optimize water usage for residential lawns using Smart Watering Systems

12 of 33

Wasted water can be reduced by optimizing the frequency and times with which watering happens, as well as optimizing how much time is spent watering different zones.

To accomplish this, computer

vision can be used to monitor the effectiveness of each watering zone, and use predictive modeling to determine the optimal quantity and frequency of watering for each zone.

Computer Vision Driven Sprinkler Optimization

13 of 33

This can be accomplished by installing video surveillance of lawns, which can utilize Computer Vision to identify areas in need of more watering.^[5]

The information gathered, in conjunction with data gathered on the effects of each watering zone can be fed to a predictive model which is able to calculate the optimal watering pattern for a given yard.

The resulting information about where water is needed is passed to a microcontroller that works with existing sprinkler controls to water the yard areas that need it at the time.

14 of 33

A Quick Summary of Computer Vision

This Computer Vision system works by being trained on large sets of data that have been labeled which allows for patterns to be identified. This can then be used to generalize to new information, such as being installed into new yards. This can then be used to build digital layouts of the size, shape, plant composition, and watering patterns of a yard.

This information can then be used in conjunction with models that are trained on observing patterns that indicate the health and need for water of plants. A prediction can then be made about the optimal settings for the given sprinkler system to best water the yard using the tools it has, such as watering zones, watering length, and frequency. We can then combine this data with information such as future weather to further optimize watering patterns to adapt to natural water that comes from rain.

15 of 33

Minimizing Cost

The biggest issue with related products is the massive cost.^[6] This is typically incurred through expensive hardware such as moisture sensors and complicated installation that can include extensive landscaping or installation of an entirely new sprinkler system.

We solve this issue by using a Computer Vision based system that is cheap and easy to install. It is able to learn the patterns of an existing system, and optimize for it without requiring any reworking of the irrigation system itself.

16 of 33

Use Cases - Why Buy This Product, Beyond Conservation?

Healthier Lawn: Optimal watering schedule based on individualized plant and grass needs
Easy Use: Hands-off process so you can go about your busy life
Peace of Mind: Automated watering when you are away from home
Money: Reduce water bill through minimizing wasted water
Compliance: The system will ensure your watering schedule complies with local regulations, so you don’t have to think about it
Continuous Improvement: Since the primary component is a software product, it can be updated and improved with no physical maintenance to the system

17 of 33

Mockup Mobile App

Two main screens:

Effective coverage to see if things have changed and updates to make
Camera monitoring and water usage stats.

September

18 of 33

Implementation & Model

19 of 33

$200

+$50

_____________

= ~$300

Cost Breakdown

*Note that Ring Camera cost already encapsulates cost of labor/software

cost of camera (based on average cost of Ring Cameras)^[17]

cost of microcontroller to connect to existing sprinkler system

miscellaneous maintenance costs

estimated cost

20 of 33

Budget For Initial Development

Item	Details	Cost
Initial Hardware:	HD Cameras, microcontrollers, wifi network, etc	$3000
Initial Software:	Swift/Kotlin native mobile apps, database connectivity, custom computer vision model, developer licenses, AWS hosting and server costs	$1000 / year
Labor Costs:	Student based development, marketing, local outreach and research.	$2000

21 of 33

Key Points

Policy

Integration

Regulation Compliance

Scalability

Opportunity for government incentives and funding (see slide 21 and 28)

Our product will integrate with existing sprinkler systems, rather than requiring installation of a new system

Algorithm will be trained to abide by local regulations regarding residential irrigation (i.e. watering times)

Works on the level of individual households, so there is unbounded scalability

22 of 33

Government Funding

The Utah Board of Water Resources currently offers funding for residential irrigation projects
“The Board’s Revolving Loan Funds provide low-interest loans to incorporated groups, such as mutual irrigation and water companies, municipalities, and other entities for many types of projects, including flood control projects. Interest rates and repayment terms are based on the Board’s affordability guidelines and the project area’s Median Adjusted Gross Income (MAGI)” ^[18]
Similar funds exist in other states and nationally, such as the Land and Water Conservation Fund
We can utilize this funding to subsidize cost

23 of 33

Timeline

Continue development and lobbying, begin advertising in Sandy, UT

Expand to communities surrounding Sandy, train algorithm for expansion (differing regulations, plants, soil types)

Product/algorithm development, lobby Utah Board of Water Resources and state representatives for funding/incentives, fundraise for initial development

Expand product to other communities vulnerable to drought (see slide 26)

Installation for first customers, evaluate and iterate on user feedback

Months 1-3

1 Year +

Months 4-6

Months 10-12

Months 7-9

24 of 33

The Impact

Reduction in suburban water consumption

25 of 33

Who Does It Affect ?

Water waste is a growing crisis, especially in drought-prone areas
Impact on communities: Increasing water costs, restrictions, and droughts
Impact on individuals: Unknowingly wasting water, higher bills, and lawn health issues
Declining water access: Locations such as The Great Salt Lake in Utah are rapidly shrinking due to evaporation and overuse, threatening wildlife, air quality, and long-term water supply

Lawns in the US cover 40 million acres, which is about the same amount of land as the entire state of Colorado!

26 of 33

Potential Water Savings by the Numbers

For homeowners: Up to 15,000 gallons saved annually per home as a direct benefit ^[14]

For national interests: 390 billion gallons saved nationwide which demonstrate the potential for large-scale community impact ^[14]

Equivalent to the annual water needs of 5 million American homes, with specific utilities offering rebates to encourage smarter irrigation practices ^[14]

Based on EPA estimates of a smart water-delivery system (see slide 27 for what sets us apart)

27 of 33

Case Study : Sandy, Utah

494.49 MILLION GALLONS SAVED (over a year)

Success in Sandy can lead to expansion into neighboring drought-prone communities and eventually nationwide.

Sandy, Utah: 32,966 households ^[20]

An estimated 30-50% of lawn water waste can be eliminated. ^[1]

^[28]

28 of 33

Precedent

GreenSight falls at the intersection of a number of existing technologies, combining and improving upon them in a new and innovative way

29 of 33

Current Systems and a Comparison

Current digital sprinkler systems have a few major problems:

Cost: Existing residential smart irrigation systems cost between $600-$3,500^[6] which make them prohibitively expensive for widespread adoption, which is required to make a notable impact
Sensor Limited: Current systems are specialized into one of two buckets, weather based and moisture sensor based. The weather based system fail to account for current plant health, and moisture based are accurate but fail to predict future weather events
Water Savings: Research shows that these systems reduce water waste by about 30-50%.^[1]

These current systems are an effective proof of concept for GreenSight, while showing room for improvement:

Cost: Our largest improvement over these systems is cost. Projecting a $300 cost before tax- incentives, and 10% water bill reduction per month. Our system would have significantly more adoption, creating statewide water savings
Using a camera with a computer vision model and an integrated app, GreenSight would rival or surpass these existing systems while being significantly cheaper
By combining the best aspects of existing systems, we project an improved water savings of 45-55%

30 of 33

Tax Incentives and Previous Policy

To promote adoption of GreenSight there should be a tax incentive or policy to subsidize installations in drought sensitive communities.

Opportunity: Drought plagued communities such as the Salt Lake Valley or Los Angeles would serve as a prime place for the initial policy implementation due to their relatively wealthy yet water conscious legislatures
Previous Policy: In 2015, Congress extended the 30% tax reimbursement on new solar installation^[23]. When coupled with reduced photovoltaic costs, this policy led to a massive rise in adoption as seen in the graph. This proves an effective precedence for similar policies.
Basic Resources: Similar to power, water is a basic resource in which there is a limited amount and no effective way to create more. In conjunction with the already low price of GreenSight, a state level tax incentive would result in greater adoption and significant water conservation

[24]

31 of 33

Existing Computer Vision in Agriculture

Large-scale

Drones and other camera based systems to identity plant health and water needs:

Computer Vision: Existing companies such as N-IX^[5]have computer vision based products implemented in drones and stationary cameras to identify plant health and water needs. By using existing optimized identification models, these algorithms would be relatively low impact in terms of water and monetary resources
Water Savings: Farms with current computer vision based systems in place have reported a roughly 40% reduction^[25] in water usage. GreenSight would likely have a similar savings even at a smaller scale as seen in previously stated research

Efficient at scale

Existing apps such as Planta and PlantSnap can identify plants, provide care tips, and diagnose diseases from a simple picture on a phone
Our product will use similar technology to determine optimal water use based on each plant species and its individual water needs
These vision models are able to run on standard phones, proving that low compute and efficient models exist, showing our product is realistic at scale

32 of 33

Thanks

We would like to say thank you to the Wilkes Climate Center for setting up this Hackathon and providing resources and amazing food during the whole process!

CREDITS: This presentation template was created by Slidesgo, including icons by Flaticon and infographics & images by Freepik

1 of 33

2 of 33