Inspiration
U.N. Sustainability Goals
- 1) No Poverty
- 2) Zero Hunger
- 3) Good Health & Well-being
- 5) Gender Equality
- 6) Clean Water & Sanitation
- 7) Affordable & Clean Energy
- 8) Decent Work & Economic Growth
- 9) Industry, Innovation & Infrastructure
- 10) Reduced Inequalities
- 11) Sustainable Cities & Communities
- 13) Cllimate Action
Problem
Agriculture, food, and related industries contributed trillions of dollars to the U.S. gross domestic product (GDP) yet farmers are failing to meet the food demands of a growing population. No Farms. No Food. No Future. In developing countries, 65% of working adults depend on agriculture for their income. POP can help end poverty, raise incomes, and improve food security building a better world for everyone. The problem we chose is the intensification of climate change effects on small-scale farmers in the developing world. We know from research that climate change threatens food security through its direct effects on crop production, as well as changes in markets and food prices. The risk to the livelihood and food security of many more smallholder farmers is set to increase. POP is a safety net to empower farmers in developing countries. It measures crops and livestock producers that optimizes the agriculture supply chain by improving yields, efficiency, and profitability. The end goal is to lift people out of poverty. Farmers are pressured to grow more with less water, fertilizers, and manual labour. Farmers who are the backbone of our society often live on the edge of poverty. Social implications includes a higher suicide rate among farmers which is 3.5 times higher than the general population. Our current food system is fragile and unsustainable failing farmers, and people everywhere. As developers, we need to step in and provide farmers with a climate-resilient solution. Achieving this environmental goal means increasing the availability of nutritious food, making food more affordable and reducing inequities in access to food. Food demand is directly correlated to the rising population and income. By 2050, the world population will be 9.1 billion, up from 7.4 billion in 2016. According to the UN, farmers must increase food production by 70 percent compared to 2007 levels to meet the needs of the larger population. Climate change will reduce farm productivity by 17%. Climate change contributes to long-term environmental problems, such as groundwater depletion and soil degradation, which greatly affects food and agriculture production systems. Currently, agriculture accounts for 70 percent of all freshwater withdrawals globally and an even higher share of “consumptive water use” due to the evapotranspiration of crops. With deforestation from agriculture leads to desertification, soil erosion, and fewer crops. Heat and water scarcity will have a direct impact on animal health and will also reduce the quality and supply of crops.
Solution
Precision Agriculture
Precision agriculture using remote sensors to create an asset tracking system that collects data on precipitation, chlorophyll, radiation, reflectance, temperature, and humidity. Computer vision technology allows farmers to save water by creating irrigation maps using extracted boundaries from satellite images. Crop detection to classify pest, diseases, and fertility by measuring colour changes, mechanical damage, nutrient deficiency, water stress, or soil compaction. Rewards for land-based carbon sequestration addressing climate change with regenerative agriculture to reverse soil degradation by preventing erosion, absorbing moisture, feeding biodiverse microbes, and increasing soil carbon levels.
On a mission to empower farmers to sustainably feed the next billion people. Farmers can now make decisions like past diagnosis price predictions an harvest scheduling with data driven analytics.
- Provide a safety net for small-scale farmers to be more autonomous.
- Scale incentives for higher resiliency against climate change.
- Improve agricultural supply chains to reduce food loss and make producer-market connections. AGRICULTURE TECHNOLOGY
- Bundled software management with an easy-to-navigate user interface that is accessible without internet with SMS texting.
- Precision agriculture using remote sensors to create an asset tracking system that collects data on precipitation, chlorophyll, radiation, reflectance, temperature, and humidity.
- Computer vision technology allows farmers to save water by creating irrigation maps using extracted boundaries from satellite images. Crop detection to classify pest, diseases, and fertility by measuring colour changes, mechanical damage, nutrient deficiency, water stress, or soil compaction.
- Rewards for land-based carbon sequestration addressing climate change with regenerative agriculture to reverse soil degradation by preventing erosion, absorbing moisture, feeding biodiverse microbes, and increasing soil carbon levels.
What it does
Technology
Platform as a Service Solution
Leverage digital technology artificial intelligence innovations with the potential applied in agriculture:
- Show potential to improve farmer incomes, productivity, and climate change resilience
- Address barriers to scaled adoption of digital artificial intelligence services, such as access, affordability and digital literacy
- Have the potential to be bundled with multiple farmer-facing services into an integrated solution enabled by robust digital and data technology platforms and services
- Provide fit-for-purpose solutions using human-centered design for farming enterprises
- Show potential to be scaled using a sustainable business model, affordable, and able to provide positive return on investment at a smallholder farm level through successful pilots, scaling partners, or higher volume production driving down prices
How we built it
Platform as a Service
AWS
- Elastic compute cloud allows data to create a virtual computer in the cloud where POP can choose operating system memory and computing power.
Example: instance as a server for web application cloud watch service can collect logs and metrics from each individual instance
*Elastic Beanstalk made that much easier by providing an additional layer of abstraction on top of EC2 Ain other auto scaling features
- Lambda came out which are functions as a service or serverless computing : simply upload POP code then choosing event that decides when that code should run
Agriculture-Vision Amazon Marketplace DataSet
AgricultureVision was accessed on DATE from https://registry.opendata.aws/intelinair_agriculture_vision .
Description Original dataset affiliated with the 2020 CVPR paper. Dataset provided as a series of tar.gz files with data for each year and an associated json file dscribing the train/validation/test split.
Resource type S3 Bucket Amazon Resource Name (ARN)
arn:aws:s3:::intelinair-data-releases/agriculture-vision/cvpr_paper_2020
AWS Region us-east-1 AWS CLI Access (No AWS account required)
aws s3 ls --no-sign-request s3://intelinair-data-releases/agriculture-vision/cvpr_paper_2020/ ' Milestone
- Merge the soil organic carbon dataset with the environmental covariates dataset in order to start training a model in GEE based on historical averages of environmental covariates.
- Conduct literature review into rates of soil carbon sequestration based on land management practice and land use type to develop a comprehensive research database of carbon sequestration rates.
- Soil carbon depends on environmental factors such as climate, land cover, and soil conditions so understanding these covariates and their impact on soil carbon will be critical to modeling future scenarios.
- Soil samples have been collected heterogeneously across time and will need to be appropriately distributed and correlated with covariate data to be useful.
- These samples can be tested against OpenLandMap to improve the quality of and error-checking of the ML approach to carbon estimation.
- Development of a research database for carbon sequestration rates by land use can then be used to test the predictive model under different future scenarios.
- Field measurements of soil, such as ISCN, WISE, and RaCA.
- Environmental covariates: EarthEnv, WorldClim2, SoilGrids, WCS, GEDI, satellite imagery, and other sources.
- Landsat data going back to 1970 with OpenLandMap data
- CGIAR based on : https://www.soilgrids.org/
- Carbon Drawdown database
- Carbon Farming Solution Milestone
- Outline the known gaps in soil organic carbon field samples and outline research priorities.
- Outline known gaps in rates of carbon sequestration under different land use scenarios and outline research priorities.
- Outline the need to understand the top-line soil organic carbon stock potential based on environmental covariates and land use scenarios.
- Select the top scenarios describing the potential for additionality over business as usual and write up a report brief for policymakers.
- Hypothesis:
- Forecast 10, 20, 30, 50 year sequestration rates using practice-based trends:
- historical trends (do nothing)
- stop all human intervention (other do nothing)
- humans intervene in the best way (best case)
- Somewhere in between (middle case) Milestone
- Develop a model based on known sequestration rates attributed to land cover, land use change, or land management practices that can describe the potential additionality for each land use over time.
- Create a map-based user interface so that users can interactively create their own scenarios based on their own assumptions of the adoption of regenerative practices.
- Naively train a baseline model that predicts depths of 30cm or 100cm, soil organic carbon stocks for each land use type we can describe the potential for additionality under different scenarios. Candidate algorithms include random forest, explainable gradient boosting machines, and TabNet.
- Through this process we also hope to identify environmental predictors for soil organic carbon growths and losses.
- Batch photo upload and labeling with Pl@ntnet dataset for plant identification where the shared dataset contains 306,293 images of 1081 species that retain a level of ambiguity allowing it to be trained for AWS computer vision precision agriculture model with detail. For example, there is a very large number of anemones and that they are often distinguished
outputs:
- US Forest biomass / carbon sequestration score
- US Above ground carbon sequestration ROI - use PLACES data set for US land valuation
- US Water restoration potential score
- US Water restoration ROI
- User can pan and zoom
- User can search for place names
- User can view data values for each point on the map, either by mousing over the map or clicking
- User can view land parcels for some predetermined rural US area
- Initial soil / below ground carbon sequestration potential data is available from the #soil team:
- Below ground carbon sequestration score
- Below ground carbon sequestration ROI
- Total sequestration score
- Total sequestration ROI
- Overall ROI
Satelite Data from Amazon Atenna
Satellite orbiting earth you can tap into Amazon's global network of antennas to connect data through its ground station service
Amazon Snow Work without Internet in hostile environments like places in Africa. Example: scientist in the Arctic snow devices are like little mini data centres
Red shift data data in a warehouse is structured so it can be queried - that tries to get you to shift away from Oracle warehouses enterprises to dump multiple data sources from the business where they can be analyzed together
Kinesis
Analyze real-time data you can use kinesis to capture real-time streams from your infrastructure then visualize them in your favorite business intelligence tool
data exchange purchase an subscribe to data from third party sources once you have some data in the cloud you can use
Amazon GLUE STUDIO ETL
- Automatically connect POP other data sources on AWS like Aurora redshift and S3
ARTIFICAL INTELLIGENCE
Make Predictions
Amazon Sage maker
to connect to it and start building machine learning models with tensorflow or pytorch it operates on multiple levels to make machine learning easier and provides a managed Jupiter notebook that can connect to a GPU instance to train a machine learning model then deploy it somewhere useful
Amazon Recognition API COMPUTER VISION building your own ML models from scratch is still extremely difficult if you need to do image analysis you may as well just use the recognition API.
Amazon Alexa NATURAL LANGUAGE PROCESSING conversational bot you might use LAX which runs on the same technology that powers Alexa devices
Remote Sensing with IOT: use Amazon Robo maker or Deep racer to simulate and test POP farming robots connected to tractors. IoT core to collect data from them update their software and manage them remotely with a a deep racer device which is an actual race car that you can drive with your own machine learning code. Predictive Maintenance. Building the demo for the tractor with the sensor, in addition to the plant irrigation system for the IOT greenhouse. Rather than waiting to make real time decisions or for farm equipment to fail, IoT sensors on the farmland can track temperature, humidity, pest/diseases, and other data points.
BLOCKCHAIN
Quantum Ledger for Carbon Credits
- Amazon Budgets with cost explorer to help ESG stakeholders and organizations plan the implementation of POP by creating rules and determine who has access POP account. Customer Insights and Product Innovation by Connecting AWS Budgets to our farm telemetry data. Innovation accelerates a farmer's ability to segment customers, customize products, set prices to better capture value, and provide value-added services. With power business intelligence can share our actionable insights discovered from our data with automated reporting. Precision agriculture using IoT we can trigger actions based off of specific conditions in your IoT setup. Example setting carbon sequestered and temperature set to a limit being sent to our email.
Challenges we ran into
Setting up Ethereum Infera Nodes
Accomplishments that we're proud of
Interested in applying blockchain technologies for good
What we learned
AWS can bridge the digital divide for farmers who need technological innovations using their easy to use interface and reporting tools we can analyze farm data like temperature, humidity, precipitation, and crop health in addition to data collected from our tractor or aerial drone this data includes accelerometer, magnetometer, and gyroscope in addition to longitude and latitude coordinates. This will be helpful for farmers deciding where to sell their food in addition to land use optimization and supply chain optimization. From our data storage pipeline we can link to custom vision and Tensorflow. This allows us to pull agriculture image data to create computer vision machine learning models. computer vision model allows farmers to remotely monitor farm land boundary patterns for the tractor in addition to crop health monitoring for the IoT greenhouse.
What's next for Proof of Plate
- Deploy a simple WordPress landing website with AWS lightsail deploying a static server that
- traffic scaling and networking with Docker containers and App Runner handles all the orchestration and scaling behind the scenes
- data warehouses to query a larger variety of data sources
- store data in the cloud glacier which has a higher latency but a much lower cost
- AWS lake formation which is a tool for creating data lakes or repository's that store a large amount of unstructured data instead of Redshift
Log in or sign up for Devpost to join the conversation.