Techathon 2

Challenge 3: Can you limit green cover depletion?


India’s first comprehensive climate analysis report published in 2020 ‘Assessment of Climate Change over the Indian region’, highlights the role of forests as effective mechanisms to mitigate climate change impact, provide economic benefits for the country and meet several of India’s sustainable development goals. Over the past two decades, India has witnessed an ever-increasing rate of deforestation and unsustainable exploitation of forest resources, leading to overall degradation at an alarming rate. Urban air quality is another area of serious health concern for society, especially with the high levels of suspended particular matter (SPM) due to construction, transportation and industrial activity. Green space has been deemed to be effective in mitigating particulate matter (PM) pollution for which civic authorities need to have reliable change data to take timely action. Measurement of green cover is difficult and contentious – and application of Artificial Intelligence (AI) to Vision Analytics combines with aerial/space photography to give us innovative solutions to enable concrete policy making and action.

Expected output

Build an innovative vision analytics solution that can analyze geo-mapped aerial/space photographs and calculate the green cover area to the best possible estimate. The solution should be able to calculate, visualize and predict the change in green cover over time. Additional relevant data layers and functionality should be used to score points for originality.

Intelligent detection of green cover and its changes should be part of the automated solution. The quality of analysis and UI that enables improved analysis like distinguishing between types of green cover, types of depletion, zoom features for manual inspection, rate of change graphs, time-lapse videos, etc. Scalable features for users to select different geographic ranges and time zones will score extra points. Multiple alternative estimation techniques to increase the confidence of measurements will also score points. Sourcing such data is a part of the challenge. Some common sources are shared below but richer/additional data sources are left to the participants.

The proposed solution user interface should preferably be through a browser (using web technologies), should support API integration, and allow cloud deployment using containerization (K8s/ Docker are encouraged).

Technology and standards

  • Python, Java, C++, a or any common programming language, and common Web application framework.
  • EY will open ASpace, our proprietary AI platform to a limited number of participants on a first come first serve basis to develop these solutions.

Reference datasets

Google Earth datasets, Public datasets from NASA provided by Landsat and Sentinel. ISRO datasets. Kaggle datasets. Amazon datasets (Registry of Open Data on AWS)

Judging criteria

A scoring model basis:

  • The scope of the problem covered in the solution
  • Novelty of the idea and innovative functionality
  • Quality of data sourced and processed
  • Quality of user interface (rich visualization, intuitiveness)
  • Quality and range of analysis available including accuracy and reliability
  • Solution design framework and use of technology
  • Value realization, end-user benefit and social impact

For any queries on EY Techathon 2 #iSolve4aBillion Challenge E-mail us.

Follow us for latest updates


Challenge 2