Inspiration
The marked difference in ecological profiles and the presence of different birds in green spaces between suburbs in cities, especially when compared to average socio-economic status. We realised that urban green spaces are heavily reliant on the tools/expertise that councils are able to afford and access, and thus decided to leverage mapping and AI to provide more information and insight to users, as it is difficult to easily access and interpret this kind of detailed information for most people.
What It Does
grapevine.ai provides instant location-based, scientific recommendations on hyperlocal flora for effective urban heat island mitigation, while providing visual maps of urban heat islands, the urban heat vulnerability index and vegetation cover, to assist users in identifying ideal regions for urban green space implementation. grapevine.ai sets up immersive visualisations of local/communal urban heat islands, with options to compare with urban vegetation cover and heat vulnerability of the area, in order to deliver customised, accurate information (based on location and level of heat) on local natives effective for urban heat mitigation, while conserving endangered native flora as a secondary benefit.
How We Built It
- Frontend: HTML with CSS and Bootstrap to make a responsive website.
- Backend: Node.js used to create the scripts for the AI components.
- Database: MongoDB for storing user profiles and historical weather data.
- Mapping: ArcGIS Online for web-based mapping software, with data derived from open source datasets on urban heat islands, heat vulnerability indices and vegetation cover from the government.
Challenges
A major challenge was developing the AI aspect of the project, as it was difficult to securely integrate OpenAI through node.js with HTML - especially as we are beginners in the field of AI and chatbots. Eventually, we had to just decide to not actually integrate the AI component, because it took way too much time and resources to try setting up a separate server to securely pass the API key. We also struggled with ArcGIS Online initially - the user interface was somewhat confusing for a beginner, and our computers struggled to process the large, complex datasets into highly detailed maps. However, with trial and error and a great deal of patience, a decently functional map was produced.
Future Directions
A functional AI application + the potential for a mobile app as well as increasing mobile-friendliness of the website.
Log in or sign up for Devpost to join the conversation.