Case study
ClovAR at MonHack
A gamified AR exploration app combining a Flutter mobile flow, a Python classifier, and browser-rendered plant models.
The idea
ClovAR was our team project for MonHack, a Monash University hackathon. We wanted to give people, especially children, a reason to explore the plants around them through a small game-like mobile experience.
The build joined three pieces: a Flutter application, a Python machine-learning service that returned plant characteristics, and a locally hosted AR.js service that rendered a matching 3D plant model.
What we proved
The useful part of the prototype was the handoff between capture, classification, and the augmented-reality result. Each technology had a narrow job instead of asking the mobile application to own the whole pipeline.
The constraint
This was built under hackathon time pressure, so the goal was a coherent end-to-end interaction rather than production depth. Classification confidence, device compatibility, model loading, and accessibility would all need more work before treating the concept as a product.
What I would revisit
I would first document the service boundaries and preserve a small test set for the classifier. On the interface side, I would add a non-AR result view so the core information remains usable on unsupported devices.