PetalScan
A university final-year team project for recognising plants and landmarks at Singapore Botanic Gardens, with an AI assistant for follow-up questions.
View the build- Year
- 2025
- Role
- University final-year team project
- Discipline
- Mobile AI prototype
- Stack
- Expo · React Native · PyTorch · FastAPI · Firebase · Google Cloud Run · OpenAI · Stripe
PetalScan combines an Expo mobile app, a 131-class MobileNetV3 image-recognition model and Petal-GPT. The prototype connects a camera scan to a predicted plant or landmark and a follow-up conversation.
A university final-year team project for recognising plants and landmarks at Singapore Botanic Gardens, with an AI assistant for follow-up questions.
What moved the work forward.
Context
We built PetalScan as our university final-year project, using Singapore Botanic Gardens as the setting for a mobile recognition experience.
Build
The mobile app connects camera and scan flows to a FastAPI service on Cloud Run. Firebase supports accounts and storage, and Petal-GPT uses the predicted label as context for follow-up questions.
Scope
The classifier recognises a fixed set of plants and landmarks. Petal-GPT provides general generated responses; the project did not establish a verified horticultural or pet-safety service.
Product, interface and moments from the build.
