T-AIM (Govt. Of Telangana State) And NASSCOM Present "The Innovation Factory"


T-AIM Innovation Factory First Edition




Context:
Telangana AI Mission (T-AIM) has been established by the Government of Telangana State under the aegis of ‘2020 Year of AI’, with a vision to launch Telangana as a global hub for Artificial Intelligence and foster social innovation.

Agriculture is a key area of focus for the state. With the intention of sourcing best in class AI solutions in the field of agriculture, T-AIM is launching Innovation Factory. It is a series of Grand Challenges wherein participants will compete to develop innovative AI Solutions for specific use-cases in Agriculture.

This is an opportunity for innovators to;
• Assist the Government of Telangana in helping farmers increase their income by working on three selected use-cases
• Showcase their AI solution development capabilities at a larger scale
• Avail an opportunity to implement a pilot within the State of Telangana, with mentorship support from Government of Telangana entities
• Win a prize of INR Four Lakhs (Winner) and INR Two Lakhs (Runner-up), per use-case

Eligibility Criteria:
• This challenge is open for Start-ups/companies/Institutions operating within India

Rules:
• The contestants/Team will be allowed to have only one entry/one account per team
• No cross-team collaborations are allowed
• No private sharing of data/code/analysis/insights is allowed
• Any misleading, incorrect, and incomplete information provided will lead to the expulsion of the contestant
• Any actions or activities to influence, and/or inappropriate communication with the Program Team and related Government officials/consultants will lead to disqualification of the contestant

Results and Winning Criteria:
• Teams will be given six (6) weeks to work on the use-case
• Teams will have to submit the results, approach and code to be presented to the jury for evaluation
• The winning criteria will be a combination of qualitative and qualitative methods to solve the problem
• The decision of the jury will be final and binding.

Ranking Criteria (Parameters and Weightage)

Approach
This consists of the description of how the problem statement was approached. Assumptions made. Understanding of problem statements. Suggestion on how the solution can be taken to next level with the use of data/algorithms.
30%

Technical
This consists of Code quality, use of data, scalability of the solution, Use of standard libraries. Algorithms used. 30%

Results
This consists of Accuracy of the solution with 2020 Yield data (+- 10%) range
40%


Key Milestones and Dates:

Registration Start Date: 28th December, 2020
Registration End Date: 24th January, 2021
Contest Start Date: 27th January, 2021
Contest End Date: 10th March, 2021
Top 10 Announcement Date: 21st March, 2021
Jury Presentation Date: 27th March, 2021
Winner Announcement Date: 30th March, 2021


Use Case 1:

Precision Farming for improving Yield
Background:
Farming is a multistage process in which majorly there are 3 growth stages. Across these stages, farmers use, depending on the natural state of climate variables, resources such as water, seeds, fertilizers, pesticides, and insecticides. The farm output is yield which is dependent on these underlying variables. Using these variables, predicting the historical yield through an ML model could give us statistical relationships between variables that affect the yield of a particular crop.

Selected Crops:
• Maize
• Groundnut
• Bengal Gram

Objectives:
• Identify variable(s) in accordance with their relevant influence on crops yield
• Recommend best practices to maximize the yield for a particular crop
• Build a Machine-Learning model to predict yield




Use Case 2:

Real-time price discovery & volume management at e-marketplaces

Background:

The Agricultural market cannot just be explained by the simple principle of supply and demand. Rather, it is an interplay of many variables that ideally should be used in the price-discovery of crops. Post-harvest, farmers generally rely on the local prices set at mandis/marketplaces to sell their crops. It has been witnessed that using historical and real-time data, machine learning models could be built to predict real-time prices and also help manage crop volume.   

Selected Crops:

  • Maize
  • Groundnut
  • Bengal Gram

Objective:

  • Build a machine learning model to discover real-time prices for identified crops across e-marketplaces and accordingly manage volumes.

 

Use Case 3:

Farmer lending using farm/output backed credit risk assessment

Background:

Lending institutions in the farming sector need standardized assessment tools to give loans to farmers. The processes implemented consider many important factors such as farm yields, hazards, crop portfolios, and geographic details. However, certain factors lead to the development of the aberrated risk profiles and, corresponding loan disbursements leading to defaults affecting both the lending institutions and the farmers. Incorporating pertinent variables and using them to create credit risk assessment models help to gauge the health of borrowers and corresponding borrowing capacities, improving the lending landscape in agriculture.

Objectives:

  • Develop a credit risk assessment model for agricultural loan portfolios
  • Segment target profiles into risk buckets to promote healthy lending