AI 에이전트를 사용하여 개발된 삼림 벌채 식별 도구

DEV Weekend Challenge: Passion Edition Submission

This is a submission for Weekend Challenge: Passion Edition

weekendchallenge.

What I Built

The project is an east-to-use application which helps user to identify deforestation in areas of interest. From the users selected area of interest, application downloads
the satellite images, generates ndvi(Normalized Difference Vegetation Index), and identifies potentially deforested locations based on calculated vegetation indices.

My goal was to evaluate the capabilities of AI agents in developing a complete application, production-ready with instructions provided by human. The project also explores the
time required to build such an application with AI-assisted software.

Demo

https://huggingface.co/spaces/sgharti/crop-health

Code

https://huggingface.co/spaces/sgharti/crop-health/tree/main

How I Built It

I developed a plan for core software architecture and directed the entire application workflow including the following:

• Describing entire application lifecycle from user input to fastapi pipeline (GEE and Snowflake).
• Described the interface specification(leaflet map, design and user input)
• Described pipeline of how system connects to GEE, generates NDVI(Normalized Difference Vegetation Index) and stores in the Snowflake.
• Described the workflow of backend-frontend synchronization to read logs from snowflakes and display it on the frontend with visualization and text explanation.

I decided to use Google AI (Antigravity with gemini) to build this application.

Prize Categories

I am applying for the Google AI (Gemini, Antigravity) and Snowflake tracks.

Team Submissions:
Shashi Gharti @shashigharti