UrbanJungle
Social plant care app with AR plant identification, care reminders, and a community marketplace.
UrbanJungle transforms plant care into a social experience, building a thriving community of urban gardeners, rare plant collectors, and sustainability enthusiasts across Canada. Users can identify any plant instantly by pointing their camera at it — the AR system recognizes over 15,000 plant species and provides detailed care instructions, toxicity warnings (important for pet owners), and growth expectations specific to Canadian climate zones. The personalized care scheduler adapts watering, fertilizing, and pruning reminders based on the user plant collection, local weather data, and the specific light conditions of their home (analyzed through a one-time room scan). The social feed lets users share their plant progress photos, ask the community for advice on struggling plants, and follow "plant influencers" who share care tips and propagation tutorials. The built-in marketplace connects local growers with buyers, supporting everything from common houseplants to rare Monstera variegata cuttings with escrow-protected transactions and built-in shipping label generation. GreenCity Collective, a Montreal-based sustainability startup, envisioned UrbanJungle as more than just a plant app — they wanted to build a movement that encourages urban dwellers to bring more nature into their lives, reduce their carbon footprint, and connect with their local plant community. The app needed to appeal equally to a grandmother caring for her first succulent and a serious collector with 200+ plants.
The Challenge
Creating an app that combines AR technology, social features, and e-commerce into a cohesive, delightful experience that appeals to both casual gardeners and serious plant collectors presented unique design and technical challenges. The AR plant identification needed to work accurately in various lighting conditions, angles, and growth stages — a plant looks very different as a seedling versus a mature specimen. The care scheduling system needed to account for hundreds of variables: plant species, pot size, soil type, indoor vs outdoor, light exposure, local climate, season, and individual plant health. The social features needed robust content moderation (people get surprisingly heated about plant care advice) while maintaining a welcoming community feel. The marketplace required secure payment processing, dispute resolution, and a reputation system that plant sellers would actually trust. Performance was critical — the AR scanning needed to work in real-time without noticeable lag, and the app needed to handle push notifications for potentially hundreds of plants per user without draining their battery. Cross-platform consistency was essential since the plant community is roughly evenly split between iOS and Android.
Our Solution
We leveraged ARKit (iOS) and ARCore (Android) for the plant identification system, with a custom TensorFlow Lite model trained on a dataset of 2 million plant images covering 15,000+ species. The model runs entirely on-device, enabling instant identification without an internet connection — important for users scanning plants at garden centers with poor connectivity. We implemented a confidence scoring system that shows the top 3 matches when the model is not fully certain, and a "community verify" feature where experienced users can confirm or correct identifications. The care scheduling engine uses a rule-based system with over 800 species-specific care profiles, combined with the OpenWeatherMap API for local weather data and a custom light estimation algorithm that uses the phone ambient light sensor and user-provided room orientation. The social feed was built with a Twitter-like infinite scroll architecture using Firebase Realtime Database for instant updates, with AI-powered content moderation that flags potentially harmful plant care advice (e.g., suggesting toxic fertilizers) before it reaches other users. The marketplace uses Stripe Connect for split payments between buyers, sellers, and the platform, with an escrow system that holds funds until the buyer confirms the plant arrived healthy. The app uses Flutter for cross-platform consistency with platform-specific native modules for AR functionality. The entire project was delivered in 7 months from kickoff to App Store approval, with a team of 5 developers and 1 UI/UX designer.
Results
Tech Stack
Want Similar Results?
Let's discuss how we can build something epic for your business.
Start Your Project