Submit a photo → AI identifies species → Water health scored → Community data grows → Science happens. What takes a marine researcher 3 hours now takes 30 seconds.
Built for the NatGeo × TNC Externship (2026)
🔗 Live Demo: ocean-lens.vercel.app 🐙 GitHub: Pratyakshya10/OceanLens
Coastal communities, divers, and fishers observe marine ecosystem changes daily — but have no way to:
- Document species sightings with AI-backed identification
- Report coral bleaching, pollution, or habitat loss systematically
- Contribute to an open dataset that institutions can actually use
- See their observations on a live global map in real time
The Tamil Nadu fisherman who notices the reef looks different has no platform. The Kashmiri wetland watcher has no tool. OceanLens changes that.
Origin story: This project began watching the Saraswati River in Indore disappear with no systematic documentation. That silence is what OceanLens is built to end.
OceanLens lets anyone with a phone submit a marine or freshwater observation. AI interprets it instantly. The community owns the data.
- Go to
/submitand upload a water body photo - Enter your name, location coordinates, depth, and field notes
- AI (Groq Llama 4 Scout Vision) identifies species and assesses water health
- Observation is saved to the open dataset with bleaching level and confidence score
- Pin appears live on the
/mapwith full AI analysis /dashboardupdates with real-time stats- Export full dataset as CSV from dashboard
| Feature | Description |
|---|---|
| 🤖 AI Species Detection | Groq Llama 4 Scout Vision identifies species, bleaching level, and water health |
| 🗺️ Live Observation Map | Real pins placed from actual lat/lng coordinates, color-coded by bleaching level |
| 📊 Real-time Dashboard | Live stats — total observations, species count, AI analyses, active locations |
| 📍 Manual Coordinates | Submit observations from any location worldwide with manual lat/lng input |
| ☁️ Cloudinary Image Storage | Uploaded images stored persistently on Cloudinary CDN |
| 📥 CSV Export | Full open dataset downloadable as CSV in one click |
| 🌊 CoralWatch Scale | Bleaching rated 0–4 per CoralWatch methodology (healthy → fully bleached) |
| 📖 Impact Story Page | Live stats embedded in mission narrative — updates as community grows |
| Layer | Technology |
|---|---|
| Frontend | React + Vite + Tailwind CSS v3 |
| Backend | Node.js + Express |
| Database | MongoDB Atlas |
| AI Vision | Groq API — Llama 4 Scout 17B Vision |
| Image Storage | Cloudinary |
| Frontend Deploy | Vercel |
| Backend Deploy | Render |
An overview of what OceanLens is
Anyone with a phone can submit. GPS auto-detects location or enter coordinates manually for remote sites.
Groq Llama 4 Scout Vision analyzes the image and returns species, bleaching level (CoralWatch 0–4), AI summary, and confidence score.Export full dataset as CSV.
Every submission appears as a colored pin. Click any pin to see the full AI analysis, image, and field notes.
Complete story on how it started
# Clone the repo
git clone https://github.com/Pratyakshya10/OceanLens.git
cd OceanLens
# Backend
cd BACKEND/backend
npm install
# Add your .env (see below)
nodemon server.js
# Frontend (new terminal)
cd FRONTEND/frontend
npm install
npm run devBackend (BACKEND/backend/.env):
PORT=5000
MONGO_URI=your_mongodb_atlas_uri
GROQ_API_KEY=your_groq_api_key
CLOUDINARY_CLOUD_NAME=your_cloud_name
CLOUDINARY_API_KEY=your_cloudinary_api_key
CLOUDINARY_API_SECRET=your_cloudinary_api_secret
Frontend (FRONTEND/frontend/.env):
VITE_API_URL=http://localhost:5000
OceanLens/
├── BACKEND/
│ └── backend/
│ ├── config/
│ │ └── db.js
│ ├── controllers/
│ │ ├── observationController.js
│ │ └── mapController.js
│ ├── middleware/
│ │ ├── upload.js
│ │ └── errorHandler.js
│ ├── models/
│ │ └── Observation.js
│ ├── routes/
│ │ ├── observations.js
│ │ └── map.js
│ ├── services/
│ │ └── aiService.js
│ └── server.js
└── FRONTEND/
└── frontend/
└── src/
├── api/
├── components/
├── pages/
│ ├── HomePage.jsx
│ ├── SubmitPage.jsx
│ ├── DashboardPage.jsx
│ ├── MapPage.jsx
│ └── StoryPage.jsx
└── main.jsx
OceanLens uses the Groq API with Llama 4 Scout Vision to:
- Identify species, organisms, or environmental subjects in water body photos
- Rate coral bleaching on the CoralWatch 0–4 scale
- Generate plain-English water health summaries
- Return confidence scores for each analysis
The AI service (aiService.js) is fully isolated — it handles both Cloudinary URLs (production) and local file paths (development) automatically.
OceanLens has documented ecological conditions across India including:
- Chilika Lake (Odisha) — cyanobacterial blooms
- Gulf of Mannar (Tamil Nadu) — coral bleaching
- Wular Lake (Kashmir) — water hyacinth invasion
- Zuari River (Goa) — industrial fish kill
- Lakshadweep — reef bleaching at depth
- Sundarbans (West Bengal) — mangrove loss
- Kahn River (Indore) — origin story, severe pollution
- Open data — every observation belongs to the community
- Local knowledge — the fisher who has worked the same reef for 30 years knows things no satellite can capture
- AI as a tool, not a gatekeeper — AI interprets observations, it does not decide who gets to make them
"I started with a dying river in Indore. I want to make sure I'm part of the reason the ocean doesn't follow."