Meeting Harvesting – the startup that is using Geospatial tech to bring finance to farmers #TheNextGeo

In a world with over 7 billion people, which is set to reach over 10 billion by the end of the century, it is easy to understand the stress and importance of one of our oldest industries – agriculture. Can we grow enough food in a sustainable fashion for the billions of people that call this planet their home? The answer is YES (read this article by Brookings or pretty much the first page when google “10 billion people“). If we can bring technology to farming at scale, improve financing and reduce food wastage, feeding 10 billion people is definitely not a problem. However, in most emerging markets where farmers still rely on livestock for tilting their farm, technology is the least of their problems, access to finance is. As GeoGeeks, we know we can use remote sensing and geospatial analysis to help farmers improve their yields but can we use the same technologies to finance them? Ruchit Garg, the founder & CEO of Harvesting believes so. Read on!

Q: Ruchit, let’s get straight into it. Harvesting inc uses remote sensing in a rather interesting way. Instead of the more “obvious” way of using remote sensing and geospatial analysis to tell farmers what crops to plant, you are using it to connect finance with farmers. Tell us more!

A: Yes indeed! We wanted to take the big data that we gather through geospatial analysis towards addressing one of the core challenges that agriculture sector or small farm holder farmers face across the world. Remember that in developing countries where there is a dire need to better productivity and therefore deployment of remote sensing technology could be imperative, the same is not possible unless you have funds to use and adopt a new and modern technology.

Harvesting - Bringing Finance to Farmers using AI

Harvesting – Bringing Finance to Farmers using AI

Therefore after spending extensive time understanding the dynamics of the agriculture sector, we were clear that we need to address the big elephant in the room – finance. And use technology to overcome this hurdle. As much as remote sensing and geospatial analysis help in farm management it also is able to give insights on the probability of farm produce far more accurately based on machine learning, variables, and historical data yields. It is this gap of not being able to assess creditworthiness and his seasonalities with regard to loan payments that have resulted in a lending deficit which we are working to bridge by assessing the small farms.

Q: A credit risk score for farmers based on remote sensing data sounds awesome! But then, how do you derive a credit risk score for a farm using remote sensing? I suppose the credit risk score depends on the local weather, type of crop, soil fertility and a lot of other variables. Could you perhaps elaborate a bit on how it works?

A: Absolutely, as you mentioned credit score depends on all those factors and more. The geo-spatial technology through Normalized Difference Vegetation Index (NDVI) and  Normalized Difference Water Index (NDWI) can be used to monitor crop growth, health, and status throughout the cropping cycle and is also indicative of soil health. The resolution of publicly available data (Landsat data at 30m or Sentinel-2 data at 10m) is getting us closer to being able to model yields even on very small holdings of land.  So taking these inputs including historical yields into consideration, and customizing variables as required by financial institutions into machine learning platform, we are able to derive at a credit assessment of small farm holders across many regions in the world.

Q: Considering banks and other financial institutions have to follow due process and have to be able to quantify their decisions, how do you get them to “trust” your algorithms?

A: Firstly it is not easy for banks or financial institutions to deploy the kind of technology and resources one has to assess specifically for agriculture portfolio. For example data scientists to assess geospatial outcomes etc, it requires specialized skills  And since it is our core expertise we are able to build that confidence, secondly we customize our products based on variables that the financial institution need. So the product is flexible to include the inputs that a financial institution may need for their decision-making process. Customization and ability to handle it from the comfort of their desktop helps them to get more involved and invested.

Harvesting Credit Risk Dashboard

Harvesting Credit Risk Dashboard

Q: When you talk about loans for farmers, you cannot ignore the land ownership records and all the issues that come along with it. How do you monitor land records at Harvesting inc?

It is definitely a challenging area and is something we focus on. We have created a Land Record Monitoring Cloud platform where we are not just collecting land records irrespective of language barriers, but also are able to get real-time information on change of ownership etc. We are able to get insights on loans taken against a particular land. We provide immediate information to banks on a change in ownership or multiple loans if at all taken against a land.

Q: Any remote sensing analysis involves the possibility for false/wrong classification. Considering that your credit score is actually going to change the life of a farmer, how do you ensure that the algorithms are accurate?

A: We do have a team of experts running continuous tests at all times to ensure that room for error or wrong classification is minimized.

Q: Harvesting is a startup that heavily utilizes remote sensing and spatial analysis. Could you tell us more about your tech stack and the programming languages that you use?

A: Our system is bit complex which is built in combination of several technology stacks and public cloud infrastructures. For remote sensing analysis, we are primarily using python on Linux servers. We have build a high-performance computing environment in the cloud where 10000s of computers runs in parallel to process data and create real-time information we need.

Q: You are already working with financial institutions in Turkey, India, Uganda, Kenya, Tanzania and Brazil. In your opinion, what would be the biggest challenge as you scale up and operate in more countries across the world?

A: The technology today allows us to spread our wings and there are no boundaries. In emerging economies that we cater to, there is still a lot of decision making that is in the hands of policymakers, bureaucracy that takes time to turnaround. Even banks and financial institutions have their set patterns and take time to adopt to newer avenues.

Q: As a startup founder (and serial entrepreneur), I am sure that there are many things that you must have learned along the way. What was the most interesting feedback that you received so far?

A: Focus on the customer and solving their pain points. They do not care about your cool technology per say. As long as we have a way to solve their pain points we are good.

Q: Where do you see Harvesting 3 years from now?

A: We do hope to see ourselves as being part of the financial inclusion drive for farmers across the emerging markets. We already are working in different parts of the world and we hope to set our footprints in every territory where there is a lack of finance for farmers and have many smallholders under banking umbrella. 

Q: You are based in Fremont, California (USA) how’s the geo-startup scene? Are there many investors specifically looking at remote sensing and other geotech related companies? Are there any local meetups/events that cater to the geo-community?

A: Like customers, investors do not care about technology like remote sensing alone. They care about if a startup is able to create something lot of customers want and willing to pay. Having said that investors to understand the potential of remote sensing and there are more and more startups and funding coming along which touches geo/remote sensing.

Q: Okay, this is a tricky one – on a scale of 1 to 10 (10 being the highest), how geoawesome do you feel today? 😉 

A: I do feel 9 on 10 as we have come a long way in terms of access of geo- spatial technology, its reach, its scope, and affordability.

Q: Any closing remarks for anyone looking to start their own geo startup?  

A: Choose your tools smartly, it’s not about tech, it is about solving a real pain of a customer who is willing to pay for your solution.

If anyone would like to reach you, what would be the best way to do so?

They can reach out to us at info@harvesting.co


The Next Geo is supported by Geovation:

https://web.archive.org/web/20200913130221if_/https://i2.wp.com/geoawesomeness.com/wp-content/uploads/2018/04/Geovation-greyw-orangeb_portrait.png?resize=300%2C200

Location is everywhere, and our mission is to expand its use in the UK’s innovation community. So we’re here to help you along your journey to success. Get on board and let’s start with your idea…

Learn more about Geovation and how they can help turn your idea into reality at geovation.uk


About The Next Geo

The Next Geo is all about discovering the people and companies that are changing the geospatial industry – unearthing their stories, discovering their products, understanding their business models and celebrating their success! You can read more about the series and the vision behind it here.

We know it takes a village, and so we are thrilled to have your feedback, suggestions, and any leads you think should be featured on The Next Geo! Share with us, and we’ll share it with the world! You can reach us at info@geoawesomeness.com or via social media 🙂

Say thanks for this article (0)
The community is supported by:
Become a sponsor
#
#Airbus #Data processing #Deep Learning #Ideas #Maxar #Optical #Science
Enhancing Satellite Imagery Readability with Super-resolution Machine Learning Models
Aleks Buczkowski 05.16.2023
AWESOME 13
#0.30m #Airbus #Construction #Drones #Environment #Featured #Maxar #Optical #Science
Bridging the Gap: How Buying Satellite and Drone Data is Evolving
Aleks Buczkowski 02.1.2024
AWESOME 5
#>100m #Atmospheric #Environment #Featured #Government #MODIS #Science
A look at the ways in which satellites are used to monitor and forecast weather
Aleks Buczkowski 07.9.2023
AWESOME 5
Next article

New EOS platform lets you run image processing tasks in browser

Most of your image analysis tasks that required ENVI or Erdas Imagine software are now available online thanks to EOS Platform. This new game-changing cloud service launched by EOS Data Analytics provides GIS professionals with a one-stop solution for search, analysis, storing, and visualization of large amounts of geospatial data.

With EOS Platform you get access to an ecosystem of four mutually integrated EOS products, which together provide a powerful toolset for geospatial analysts. Image data is stored in cloud-based EOS Storage and is available for image processing or remote sensing analysis at any time; this can be a raw user file, imagery obtained from LandViewer or an output file from EOS Processing.

A list of EOS Processing workflows that can be filtered by industry, input data type

There are at least two reasons why image processing is the platform’s major asset: the processing of large data amounts runs online and offers as many as 16 workflows with even more coming soon. On top of that, users can get the best cartographic features of EOS Vision for vector data visualization and, as announced for the future, its analysis.

Data agnostic platform

When it comes to raster data, you can work with a variety of satellite and airborne datasets in LandViewer, EOS Processing, and EOS Storage. Users can also upload their own GeoTiff, JPEG, JPEG 2000 files and apply GIS data processing algorithms via API or from the web interface. EOS Vision is your tool for vector data operations with multiple formats support (ESRI Shapefile, GeoJSON, KML, KMZ).

The whole package for image processing

EOS Processing offers a great experience with its sixteen processing workflows, including the popular raster tools (merge, reprojection, pansharpening), remote sensing analytics, photogrammetry, and proprietary feature extraction algorithms that can’t be found anywhere else. Get your data ready for the upcoming LiDAR analysis and 3D modeling as they’ll become available soon.

Such pre-processing tasks as cloud detection or radiometric calibration help you refine raw data for further analysis: you can correct images for atmospheric effects and obtain the real ground radiance or reflectance values. 

BEFORE: A high spatial resolution image of a densely built-up area

AFTER: A grayscale image that shows buildings detected using the neural network classification algorithm

Object detection, change detection, and classification

The convolutional neural networks, pre-trained by EOS Data Analytics to extract features from imagery, let you apply state-of-art methods to detect objects and track changes from space.

  • Having only a set of multi-temporal images and change detection workflow, you can track how illegal deforestation progresses over time.
  • Edge detection can show the exact boundaries of your agricultural lands down to the last pixel.
  • It is possible to estimate the parking lot traffic of largest shopping centers with car detection algorithm.

The best of spectral analysis

Products within EOS Platform support almost all remote sensor types, and the user can choose from a long list of spectral indices to calculate on the fly. Aside from the complete set of vegetation indices (NDVI, ReCI, ARVI, SAVI, AVI, etc.), there are also indices to outline landscape features (water, snow, and ice – NDWI, NDSI) and burned areas (NBR). The greatest thing is that here you get the freedom of experimenting with spectral bands and can create custom band combinations that best fit your purposes.

BEFORE: Agricultural lands in Southern Ukraine

AFTER: A raster layer of fields’ edges precisely detected by EOS proprietary algorithm

Customize and analyze

The user-friendly interface of EOS Processing makes it easy to manage processing workflows depending on the user’s business needs. You can set the parameters for processing and repeatedly use such customized workflow to automate high-frequency analytical tasks. The coming updates will add an ability to create custom algorithms from the available data processing operations.

Agriculture, forestry, oil and gas, and more industries

A tandem of EOS products offers a much-needed solution for individuals, businesses, and organizations across numerous industries.

With vegetation indices and crop classification feature, agronomists can continuously monitor crop conditions to detect plant diseases, pests, droughts. Forestry specialists can assess fire damages, monitor forest health, track and enforce logging restrictions.

EOS Platform is a great choice for regional and urban planning, helping users identify land cover classes to generate a vegetation map. It can also make a complete list of urban features like buildings, roads, and other major features in the region.

The platform can tackle disaster management by measuring flood extent and finding fire boundaries. When it comes to oil and gas, it is capable of identifying oil rigs and assessing the environmental impact.

Classification of coniferous vs. deciduous forests in Ukraine using Sentinel-2 image

EOS Data Analytics uses cloud-based services to address different verticals with a single platform, scientifically proven analytics, scale-ups and it builds products that could add value to remotely sensed data to deliver expert-level results for your business.

Unlock the full potential of Earth observation data with EOS Platform, directly in your browser: https://eos.com/platform

Read on
Search